languages/LanguageConverter.php

   1 <?php
   2 /**
   3  * Contains the LanguageConverter class and ConverterRule class
   4  *
   5  * This program is free software; you can redistribute it and/or modify
   6  * it under the terms of the GNU General Public License as published by
   7  * the Free Software Foundation; either version 2 of the License, or
   8  * (at your option) any later version.
   9  *
  10  * This program is distributed in the hope that it will be useful,
  11  * but WITHOUT ANY WARRANTY; without even the implied warranty of
  12  * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the
  13  * GNU General Public License for more details.
  14  *
  15  * You should have received a copy of the GNU General Public License along
  16  * with this program; if not, write to the Free Software Foundation, Inc.,
  17  * 51 Franklin Street, Fifth Floor, Boston, MA 02110-1301, USA.
  18  * http://www.gnu.org/copyleft/gpl.html
  19  *
  20  * @file
  21  * @ingroup Language
  22  */
  23
  24 /**
  25  * Base class for language conversion.
  26  * @ingroup Language
  27  *
  28  * @author Zhengzhu Feng <zhengzhu@gmail.com>
  29  * @maintainers fdcn <fdcn64@gmail.com>, shinjiman <shinjiman@gmail.com>, PhiLiP <philip.npc@gmail.com>
  30  */
  31 class LanguageConverter {
  32         var $mMainLanguageCode;
  33         var $mVariants, $mVariantFallbacks, $mVariantNames;
  34         var $mTablesLoaded = false;
  35         var $mTables;
  36         // 'bidirectional' 'unidirectional' 'disable' for each variant
  37         var $mManualLevel;
  38         var $mCacheKey;
  39         var $mLangObj;
  40         var $mFlags;
  41         var $mDescCodeSep = ':', $mDescVarSep = ';';
  42         var $mUcfirst = false;
  43         var $mConvRuleTitle = false;
  44         var $mURLVariant;
  45         var $mUserVariant;
  46         var $mHeaderVariant;
  47         var $mMaxDepth = 10;
  48         var $mVarSeparatorPattern;
  49
  50         const CACHE_VERSION_KEY = 'VERSION 6';
  51
  52         /**
  53          * Constructor
  54          *
  55          * @param $langobj The Language Object
  56          * @param $maincode String: the main language code of this language
  57          * @param $variants Array: the supported variants of this language
  58          * @param $variantfallbacks Array: the fallback language of each variant
  59          * @param $flags Array: defining the custom strings that maps to the flags
  60          * @param $manualLevel Array: limit for supported variants
  61          */
  62         public function __construct( $langobj, $maincode,
  63                                                                 $variants = array(),
  64                                                                 $variantfallbacks = array(),
  65                                                                 $flags = array(),
  66                                                                 $manualLevel = array() ) {
  67                 global $wgDisabledVariants, $wgLanguageNames;
  68                 $this->mLangObj = $langobj;
  69                 $this->mMainLanguageCode = $maincode;
  70                 $this->mVariants = array_diff( $variants, $wgDisabledVariants );
  71                 $this->mVariantFallbacks = $variantfallbacks;
  72                 $this->mVariantNames = $wgLanguageNames;
  73                 $this->mCacheKey = wfMemcKey( 'conversiontables', $maincode );
  74                 $defaultflags = array(
  75                         // 'S' show converted text
  76                         // '+' add rules for alltext
  77                         // 'E' the gave flags is error
  78                         // these flags above are reserved for program
  79                         'A' => 'A',       // add rule for convert code (all text convert)
  80                         'T' => 'T',       // title convert
  81                         'R' => 'R',       // raw content
  82                         'D' => 'D',       // convert description (subclass implement)
  83                         '-' => '-',       // remove convert (not implement)
  84                         'H' => 'H',       // add rule for convert code
  85                                                   // (but no display in placed code )
  86                         'N' => 'N'        // current variant name
  87                 );
  88                 $this->mFlags = array_merge( $defaultflags, $flags );
  89                 foreach ( $this->mVariants as $v ) {
  90                         if ( array_key_exists( $v, $manualLevel ) ) {
  91                                 $this->mManualLevel[$v] = $manualLevel[$v];
  92                         } else {
  93                                 $this->mManualLevel[$v] = 'bidirectional';
  94                         }
  95                         $this->mFlags[$v] = $v;
  96                 }
  97         }
  98
  99         /**
 100          * Get all valid variants.
 101          * Call this instead of using $this->mVariants directly.
 102          *
 103          * @return Array: contains all valid variants
 104          */
 105         public function getVariants() {
 106                 return $this->mVariants;
 107         }
 108
 109         /**
 110          * In case some variant is not defined in the markup, we need
 111          * to have some fallback. For example, in zh, normally people
 112          * will define zh-hans and zh-hant, but less so for zh-sg or zh-hk.
 113          * when zh-sg is preferred but not defined, we will pick zh-hans
 114          * in this case. Right now this is only used by zh.
 115          *
 116          * @param $variant String: the language code of the variant
 117          * @return String: The code of the fallback language or the
 118          *                               main code if there is no fallback
 119          */
 120         public function getVariantFallbacks( $variant ) {
 121                 if ( isset( $this->mVariantFallbacks[$variant] ) ) {
 122                         return $this->mVariantFallbacks[$variant];
 123                 }
 124                 return $this->mMainLanguageCode;
 125         }
 126
 127         /**
 128          * Get the title produced by the conversion rule.
 129          * @return String: The converted title text
 130          */
 131         public function getConvRuleTitle() {
 132                 return $this->mConvRuleTitle;
 133         }
 134
 135         /**
 136          * Get preferred language variant.
 137          * @param $fromUser Boolean: get it from $wgUser's preferences
 138          * @param $fromHeader Boolean: get it from Accept-Language
 139          * @return String: the preferred language code
 140          */
 141         public function getPreferredVariant( $fromUser = true, $fromHeader = false ) {
 142                 global $wgDefaultLanguageVariant;
 143
 144                 $req = $this->getURLVariant();
 145
 146                 if ( $fromUser && !$req ) {
 147                         $req = $this->getUserVariant();
 148                 }
 149
 150                 if ( $fromHeader && !$req ) {
 151                         $req = $this->getHeaderVariant();
 152                 }
 153
 154                 if ( $wgDefaultLanguageVariant && !$req ) {
 155                         $req = $this->validateVariant( $wgDefaultLanguageVariant );
 156                 }
 157
 158                 // This function, unlike the other get*Variant functions, is
 159                 // not memoized (i.e. there return value is not cached) since
 160                 // new information might appear during processing after this
 161                 // is first called.
 162                 if ( $req ) {
 163                         return $req;
 164                 }
 165                 return $this->mMainLanguageCode;
 166         }
 167
 168         /**
 169          * Validate the variant
 170          * @param $variant String: the variant to validate
 171          * @return Mixed: returns the variant if it is valid, null otherwise
 172          */
 173         protected function validateVariant( $variant = null ) {
 174                 if ( $variant !== null &&
 175                          in_array( $variant, $this->mVariants ) ) {
 176                         return $variant;
 177                 }
 178                 return null;
 179         }
 180
 181         /**
 182          * Get the variant specified in the URL
 183          *
 184          * @return Mixed: variant if one found, false otherwise.
 185          */
 186         protected function getURLVariant() {
 187                 global $wgRequest;
 188
 189                 if ( $this->mURLVariant ) {
 190                         return $this->mURLVariant;
 191                 }
 192
 193                 // see if the preference is set in the request
 194                 $ret = $wgRequest->getText( 'variant' );
 195
 196                 if ( !$ret ) {
 197                         $ret = $wgRequest->getVal( 'uselang' );
 198                 }
 199
 200                 return $this->mURLVariant = $this->validateVariant( $ret );
 201         }
 202
 203         /**
 204          * Determine if the user has a variant set.
 205          *
 206          * @return Mixed: variant if one found, false otherwise.
 207          */
 208         protected function getUserVariant() {
 209                 global $wgUser;
 210
 211                 // memoizing this function wreaks havoc on parserTest.php
 212                 /* if ( $this->mUserVariant ) { */
 213                 /*      return $this->mUserVariant; */
 214                 /* } */
 215
 216                 // get language variant preference from logged in users
 217                 // Don't call this on stub objects because that causes infinite
 218                 // recursion during initialisation
 219                 if ( $wgUser->isLoggedIn() )  {
 220                         $ret = $wgUser->getOption( 'variant' );
 221                 }
 222                 else {
 223                         // figure out user lang without constructing wgLang to avoid
 224                         // infinite recursion
 225                         $ret = $wgUser->getOption( 'language' );
 226                 }
 227
 228                 return $this->mUserVariant = $this->validateVariant( $ret );
 229         }
 230
 231         /**
 232          * Determine the language variant from the Accept-Language header.
 233          *
 234          * @return Mixed: variant if one found, false otherwise.
 235          */
 236         protected function getHeaderVariant() {
 237                 global $wgRequest;
 238
 239                 if ( $this->mHeaderVariant ) {
 240                         return $this->mHeaderVariant;
 241                 }
 242
 243                 // see if some supported language variant is set in the
 244                 // http header.
 245                 $languages = array_keys( $wgRequest->getAcceptLang() );
 246                 if ( empty( $languages ) ) {
 247                         return null;
 248                 }
 249
 250                 $fallback_languages = array();
 251                 foreach ( $languages as $language ) {
 252                         $this->mHeaderVariant = $this->validateVariant( $language );
 253                         if ( $this->mHeaderVariant ) {
 254                                 break;
 255                         }
 256
 257                         // To see if there are fallbacks of current language.
 258                         // We record these fallback variants, and process
 259                         // them later.
 260                         $fallbacks = $this->getVariantFallbacks( $language );
 261                         if ( is_string( $fallbacks ) ) {
 262                                 $fallback_languages[] = $fallbacks;
 263                         } elseif ( is_array( $fallbacks ) ) {
 264                                 $fallback_languages =
 265                                         array_merge( $fallback_languages,
 266                                                                  $fallbacks );
 267                         }
 268                 }
 269
 270                 if ( !$this->mHeaderVariant ) {
 271                         // process fallback languages now
 272                         $fallback_languages = array_unique( $fallback_languages );
 273                         foreach ( $fallback_languages as $language ) {
 274                                 $this->mHeaderVariant = $this->validateVariant( $language );
 275                                 if ( $this->mHeaderVariant ) {
 276                                         break;
 277                                 }
 278                         }
 279                 }
 280
 281                 return $this->mHeaderVariant;
 282         }
 283
 284         /**
 285          * Caption convert, base on preg_replace_callback.
 286          *
 287          * To convert text in "title" or "alt", like '<img alt="text" ... '
 288          * or '<span title="text" ... '
 289          *
 290          * @return String like ' alt="yyyy"' or ' title="yyyy"'
 291          */
 292         protected function captionConvert( $matches ) {
 293                 $toVariant = $this->getPreferredVariant();
 294                 $title = $matches[1];
 295                 $text  = $matches[2];
 296                 // we convert captions except URL
 297                 if ( !strpos( $text, '://' ) ) {
 298                         $text = $this->translate( $text, $toVariant );
 299                 }
 300                 return " $title=\"$text\"";
 301         }
 302
 303         /**
 304          * Dictionary-based conversion.
 305          * This function would not parse the conversion rules.
 306          * If you want to parse rules, try to use convert() or
 307          * convertTo().
 308          *
 309          * @param $text String: the text to be converted
 310          * @param $toVariant String: the target language code
 311          * @return String: the converted text
 312          */
 313         public function autoConvert( $text, $toVariant = false ) {
 314                 wfProfileIn( __METHOD__ );
 315
 316                 $this->loadTables();
 317
 318                 if ( !$toVariant ) {
 319                         $toVariant = $this->getPreferredVariant();
 320                         if ( !$toVariant ) {
 321                                 return $text;
 322                         }
 323                 }
 324
 325                 /* we convert everything except:
 326                    1. html markups (anything between < and >)
 327                    2. html entities
 328                    3. place holders created by the parser
 329                 */
 330                 global $wgParser;
 331                 if ( isset( $wgParser ) && $wgParser->UniqPrefix() != '' ) {
 332                         $marker = '|' . $wgParser->UniqPrefix() . '[\-a-zA-Z0-9]+';
 333                 } else {
 334                         $marker = '';
 335                 }
 336
 337                 // this one is needed when the text is inside an html markup
 338                 $htmlfix = '|<[^>]+$|^[^<>]*>';
 339
 340                 // disable convert to variants between <code></code> tags
 341                 $codefix = '<code>.+?<\/code>|';
 342                 // disable convertsion of <script type="text/javascript"> ... </script>
 343                 $scriptfix = '<script.*?>.*?<\/script>|';
 344                 // disable conversion of <pre xxxx> ... </pre>
 345                 $prefix = '<pre.*?>.*?<\/pre>|';
 346
 347                 $reg = '/' . $codefix . $scriptfix . $prefix .
 348                         '<[^>]+>|&[a-zA-Z#][a-z0-9]+;' . $marker . $htmlfix . '/s';
 349
 350                 $matches = preg_split( $reg, $text, - 1, PREG_SPLIT_OFFSET_CAPTURE );
 351
 352                 $m = array_shift( $matches );
 353
 354                 $ret = $this->translate( $m[0], $toVariant );
 355                 $mstart = $m[1] + strlen( $m[0] );
 356
 357                 // enable convertsion of '<img alt="xxxx" ... '
 358                 // or '<span title="xxxx" ... '
 359                 $captionpattern  = '/\s(title|alt)\s*=\s*"([\s\S]*?)"/';
 360
 361                 $trtext = '';
 362                 $trtextmark = "\0";
 363                 $notrtext = array();
 364                 foreach ( $matches as $m ) {
 365                         $mark = substr( $text, $mstart, $m[1] - $mstart );
 366                         $mark = preg_replace_callback( $captionpattern,
 367                                                                                    array( &$this, 'captionConvert' ),
 368                                                                                    $mark );
 369                         // Let's convert the trtext only once,
 370                         // it would give us more performance improvement
 371                         $notrtext[] = $mark;
 372                         $trtext .= $m[0] . $trtextmark;
 373                         $mstart = $m[1] + strlen( $m[0] );
 374                 }
 375                 $notrtext[] = '';
 376                 $trtext = $this->translate( $trtext, $toVariant );
 377                 $trtext = StringUtils::explode( $trtextmark, $trtext );
 378                 foreach ( $trtext as $t ) {
 379                         $ret .= array_shift( $notrtext );
 380                         $ret .= $t;
 381                 }
 382                 wfProfileOut( __METHOD__ );
 383                 return $ret;
 384         }
 385
 386         /**
 387          * Translate a string to a variant.
 388          * Doesn't parse rules or do any of that other stuff, for that use
 389          * convert() or convertTo().
 390          *
 391          * @param $text String: text to convert
 392          * @param $variant String: variant language code
 393          * @return String: translated text
 394          */
 395         protected function translate( $text, $variant ) {
 396                 wfProfileIn( __METHOD__ );
 397                 // If $text is empty or only includes spaces, do nothing
 398                 // Otherwise translate it
 399                 if ( trim( $text ) ) {
 400                         $this->loadTables();
 401                         $text = $this->mTables[$variant]->replace( $text );
 402                 }
 403                 wfProfileOut( __METHOD__ );
 404                 return $text;
 405         }
 406
 407         /**
 408          * Call translate() to convert text to all valid variants.
 409          *
 410          * @param $text String: the text to be converted
 411          * @return Array: variant => converted text
 412          */
 413         public function autoConvertToAllVariants( $text ) {
 414                 wfProfileIn( __METHOD__ );
 415                 $this->loadTables();
 416
 417                 $ret = array();
 418                 foreach ( $this->mVariants as $variant ) {
 419                         $ret[$variant] = $this->translate( $text, $variant );
 420                 }
 421
 422                 wfProfileOut( __METHOD__ );
 423                 return $ret;
 424         }
 425
 426         /**
 427          * Convert link text to all valid variants.
 428          * In the first, this function only convert text outside the
 429          * "-{" "}-" markups. Since the "{" and "}" are not allowed in
 430          * titles, the text will get all converted always.
 431          * So I removed this feature and deprecated the function.
 432          *
 433          * @param $text String: the text to be converted
 434          * @return Array: variant => converted text
 435          * @deprecated Use autoConvertToAllVariants() instead
 436          */
 437         public function convertLinkToAllVariants( $text ) {
 438                 return $this->autoConvertToAllVariants( $text );
 439         }
 440
 441         /**
 442          * Apply manual conversion rules.
 443          *
 444          * @param $convRule Object: Object of ConverterRule
 445          */
 446         protected function applyManualConv( $convRule ) {
 447                 // Use syntax -{T|zh-cn:TitleCN; zh-tw:TitleTw}- to custom
 448                 // title conversion.
 449                 // Bug 24072: $mConvRuleTitle was overwritten by other manual
 450                 // rule(s) not for title, this breaks the title conversion.
 451                 $newConvRuleTitle = $convRule->getTitle();
 452                 if ( $newConvRuleTitle ) {
 453                         // So I add an empty check for getTitle()
 454                         $this->mConvRuleTitle = $newConvRuleTitle;
 455                 }
 456
 457                 // merge/remove manual conversion rules to/from global table
 458                 $convTable = $convRule->getConvTable();
 459                 $action = $convRule->getRulesAction();
 460                 foreach ( $convTable as $variant => $pair ) {
 461                         if ( !$this->validateVariant( $variant ) ) {
 462                                 continue;
 463                         }
 464
 465                         if ( $action == 'add' ) {
 466                                 foreach ( $pair as $from => $to ) {
 467                                         // to ensure that $from and $to not be left blank
 468                                         // so $this->translate() could always return a string
 469                                         if ( $from || $to ) {
 470                                                 // more efficient than array_merge(), about 2.5 times.
 471                                                 $this->mTables[$variant]->setPair( $from, $to );
 472                                         }
 473                                 }
 474                         } elseif ( $action == 'remove' ) {
 475                                 $this->mTables[$variant]->removeArray( $pair );
 476                         }
 477                 }
 478         }
 479
 480         /**
 481          * Auto convert a Title object to a readable string in the
 482          * preferred variant.
 483          *
 484          *@param $title Object: a object of Title
 485          *@return String: converted title text
 486          */
 487         public function convertTitle( $title ) {
 488                 $variant = $this->getPreferredVariant();
 489                 $index = $title->getNamespace();
 490                 if ( $index === NS_MAIN ) {
 491                         $text = '';
 492                 } else {
 493                         // first let's check if a message has given us a converted name
 494                         $nsConvKey = 'conversion-ns' . $index;
 495                         if ( !wfEmptyMsg( $nsConvKey ) ) {
 496                                 $text = wfMsgForContentNoTrans( $nsConvKey );
 497                         } else {
 498                                 // the message does not exist, try retrieve it from the current
 499                                 // variant's namespace names.
 500                                 $langObj = $this->mLangObj->factory( $variant );
 501                                 $text = $langObj->getFormattedNsText( $index );
 502                         }
 503                         $text .= ':';
 504                 }
 505                 $text .= $title->getText();
 506                 $text = $this->translate( $text, $variant );
 507                 return $text;
 508         }
 509
 510         /**
 511          * Convert text to different variants of a language. The automatic
 512          * conversion is done in autoConvert(). Here we parse the text
 513          * marked with -{}-, which specifies special conversions of the
 514          * text that can not be accomplished in autoConvert().
 515          *
 516          * Syntax of the markup:
 517          * -{code1:text1;code2:text2;...}-  or
 518          * -{flags|code1:text1;code2:text2;...}-  or
 519          * -{text}- in which case no conversion should take place for text
 520          *
 521          * @param $text String: text to be converted
 522          * @return String: converted text
 523          */
 524         public function convert( $text ) {
 525                 $variant = $this->getPreferredVariant();
 526                 return $this->convertTo( $text, $variant );
 527         }
 528
 529         /**
 530          * Same as convert() except a extra parameter to custom variant.
 531          *
 532          * @param $text String: text to be converted
 533          * @param $variant String: the target variant code
 534          * @return String: converted text
 535          */
 536         public function convertTo( $text, $variant ) {
 537                 global $wgDisableLangConversion;
 538                 if ( $wgDisableLangConversion ) return $text;
 539                 return $this->recursiveConvertTopLevel( $text, $variant );
 540         }
 541
 542         /**
 543          * Recursively convert text on the outside. Allow to use nested
 544          * markups to custom rules.
 545          *
 546          * @param $text String: text to be converted
 547          * @param $variant String: the target variant code
 548          * @param $depth Integer: depth of recursion
 549          * @return String: converted text
 550          */
 551         protected function recursiveConvertTopLevel( $text, $variant, $depth = 0 ) {
 552                 $startPos = 0;
 553                 $out = '';
 554                 $length = strlen( $text );
 555                 while ( $startPos < $length ) {
 556                         $pos = strpos( $text, '-{', $startPos );
 557
 558                         if ( $pos === false ) {
 559                                 // No more markup, append final segment
 560                                 $out .= $this->autoConvert( substr( $text, $startPos ), $variant );
 561                                 return $out;
 562                         }
 563
 564                         // Markup found
 565                         // Append initial segment
 566                         $out .= $this->autoConvert( substr( $text, $startPos, $pos - $startPos ), $variant );
 567
 568                         // Advance position
 569                         $startPos = $pos;
 570
 571                         // Do recursive conversion
 572                         $out .= $this->recursiveConvertRule( $text, $variant, $startPos, $depth + 1 );
 573                 }
 574
 575                 return $out;
 576         }
 577
 578         /**
 579          * Recursively convert text on the inside.
 580          *
 581          * @param $text String: text to be converted
 582          * @param $variant String: the target variant code
 583          * @param $depth Integer: depth of recursion
 584          * @return String: converted text
 585          */
 586         protected function recursiveConvertRule( $text, $variant, &$startPos, $depth = 0 ) {
 587                 // Quick sanity check (no function calls)
 588                 if ( $text[$startPos] !== '-' || $text[$startPos + 1] !== '{' ) {
 589                         throw new MWException( __METHOD__ . ': invalid input string' );
 590                 }
 591
 592                 $startPos += 2;
 593                 $inner = '';
 594                 $warningDone = false;
 595                 $length = strlen( $text );
 596
 597                 while ( $startPos < $length ) {
 598                         $m = false;
 599                         preg_match( '/-\{|\}-/', $text, $m,  PREG_OFFSET_CAPTURE, $startPos );
 600                         if ( !$m ) {
 601                                 // Unclosed rule
 602                                 break;
 603                         }
 604
 605                         $token = $m[0][0];
 606                         $pos = $m[0][1];
 607
 608                         // Markup found
 609                         // Append initial segment
 610                         $inner .= substr( $text, $startPos, $pos - $startPos );
 611
 612                         // Advance position
 613                         $startPos = $pos;
 614
 615                         switch ( $token ) {
 616                                 case '-{':
 617                                         // Check max depth
 618                                         if ( $depth >= $this->mMaxDepth ) {
 619                                                 $inner .= '-{';
 620                                                 if ( !$warningDone ) {
 621                                                         $inner .= '<span class="error">' .
 622                                                                 wfMsgForContent( 'language-converter-depth-warning',
 623                                                                         $this->mMaxDepth ) .
 624                                                                 '</span>';
 625                                                         $warningDone = true;
 626                                                 }
 627                                                 $startPos += 2;
 628                                                 continue;
 629                                         }
 630                                         // Recursively parse another rule
 631                                         $inner .= $this->recursiveConvertRule( $text, $variant, $startPos, $depth + 1 );
 632                                         break;
 633                                 case '}-':
 634                                         // Apply the rule
 635                                         $startPos += 2;
 636                                         $rule = new ConverterRule( $inner, $this );
 637                                         $rule->parse( $variant );
 638                                         $this->applyManualConv( $rule );
 639                                         return $rule->getDisplay();
 640                                 default:
 641                                         throw new MWException( __METHOD__ . ': invalid regex match' );
 642                         }
 643                 }
 644
 645                 // Unclosed rule
 646                 if ( $startPos < $length ) {
 647                         $inner .= substr( $text, $startPos );
 648                 }
 649                 $startPos = $length;
 650                 return '-{' . $this->autoConvert( $inner, $variant );
 651         }
 652
 653         /**
 654          * If a language supports multiple variants, it is
 655          * possible that non-existing link in one variant
 656          * actually exists in another variant. This function
 657          * tries to find it. See e.g. LanguageZh.php
 658          *
 659          * @param $link String: the name of the link
 660          * @param $nt Mixed: the title object of the link
 661          * @param $ignoreOtherCond Boolean: to disable other conditions when
 662          *              we need to transclude a template or update a category's link
 663          * @return Null, the input parameters may be modified upon return
 664          */
 665         public function findVariantLink( &$link, &$nt, $ignoreOtherCond = false ) {
 666                 # If the article has already existed, there is no need to
 667                 # check it again, otherwise it may cause a fault.
 668                 if ( is_object( $nt ) && $nt->exists() ) {
 669                         return;
 670                 }
 671
 672                 global $wgDisableLangConversion, $wgDisableTitleConversion, $wgRequest,
 673                         $wgUser;
 674                 $isredir = $wgRequest->getText( 'redirect', 'yes' );
 675                 $action = $wgRequest->getText( 'action' );
 676                 $linkconvert = $wgRequest->getText( 'linkconvert', 'yes' );
 677                 $disableLinkConversion = $wgDisableLangConversion
 678                         || $wgDisableTitleConversion;
 679                 $linkBatch = new LinkBatch();
 680
 681                 $ns = NS_MAIN;
 682
 683                 if ( $disableLinkConversion ||
 684                          ( !$ignoreOtherCond &&
 685                            ( $isredir == 'no'
 686                                  || $action == 'edit'
 687                                  || $action == 'submit'
 688                                  || $linkconvert == 'no'
 689                                  || $wgUser->getOption( 'noconvertlink' ) == 1 ) ) ) {
 690                         return;
 691                 }
 692
 693                 if ( is_object( $nt ) ) {
 694                         $ns = $nt->getNamespace();
 695                 }
 696
 697                 $variants = $this->autoConvertToAllVariants( $link );
 698                 if ( !$variants ) { // give up
 699                         return;
 700                 }
 701
 702                 $titles = array();
 703
 704                 foreach ( $variants as $v ) {
 705                         if ( $v != $link ) {
 706                                 $varnt = Title::newFromText( $v, $ns );
 707                                 if ( !is_null( $varnt ) ) {
 708                                         $linkBatch->addObj( $varnt );
 709                                         $titles[] = $varnt;
 710                                 }
 711                         }
 712                 }
 713
 714                 // fetch all variants in single query
 715                 $linkBatch->execute();
 716
 717                 foreach ( $titles as $varnt ) {
 718                         if ( $varnt->getArticleID() > 0 ) {
 719                                 $nt = $varnt;
 720                                 $link = $varnt->getText();
 721                                 break;
 722                         }
 723                 }
 724         }
 725
 726         /**
 727          * Returns language specific hash options.
 728          */
 729         public function getExtraHashOptions() {
 730                 $variant = $this->getPreferredVariant();
 731                 return '!' . $variant ;
 732         }
 733
 734         /**
 735          * Load default conversion tables.
 736          * This method must be implemented in derived class.
 737          *
 738          * @private
 739          */
 740         function loadDefaultTables() {
 741                 $name = get_class( $this );
 742                 wfDie( "Must implement loadDefaultTables() method in class $name" );
 743         }
 744
 745         /**
 746          * Load conversion tables either from the cache or the disk.
 747          * @private
 748          */
 749         function loadTables( $fromcache = true ) {
 750                 if ( $this->mTablesLoaded ) {
 751                         return;
 752                 }
 753                 global $wgMemc;
 754                 wfProfileIn( __METHOD__ );
 755                 $this->mTablesLoaded = true;
 756                 $this->mTables = false;
 757                 if ( $fromcache ) {
 758                         wfProfileIn( __METHOD__ . '-cache' );
 759                         $this->mTables = $wgMemc->get( $this->mCacheKey );
 760                         wfProfileOut( __METHOD__ . '-cache' );
 761                 }
 762                 if ( !$this->mTables
 763                          || !array_key_exists( self::CACHE_VERSION_KEY, $this->mTables ) ) {
 764                         wfProfileIn( __METHOD__ . '-recache' );
 765                         // not in cache, or we need a fresh reload.
 766                         // we will first load the default tables
 767                         // then update them using things in MediaWiki:Zhconversiontable/*
 768                         $this->loadDefaultTables();
 769                         foreach ( $this->mVariants as $var ) {
 770                                 $cached = $this->parseCachedTable( $var );
 771                                 $this->mTables[$var]->mergeArray( $cached );
 772                         }
 773
 774                         $this->postLoadTables();
 775                         $this->mTables[self::CACHE_VERSION_KEY] = true;
 776
 777                         $wgMemc->set( $this->mCacheKey, $this->mTables, 43200 );
 778                         wfProfileOut( __METHOD__ . '-recache' );
 779                 }
 780                 wfProfileOut( __METHOD__ );
 781         }
 782
 783         /**
 784          * Hook for post processig after conversion tables are loaded.
 785          *
 786          */
 787         function postLoadTables() { }
 788
 789         /**
 790          * Reload the conversion tables.
 791          *
 792          * @private
 793          */
 794         function reloadTables() {
 795                 if ( $this->mTables ) {
 796                         unset( $this->mTables );
 797                 }
 798                 $this->mTablesLoaded = false;
 799                 $this->loadTables( false );
 800         }
 801
 802
 803         /**
 804          * Parse the conversion table stored in the cache.
 805          *
 806          * The tables should be in blocks of the following form:
 807          *              -{
 808          *                      word => word ;
 809          *                      word => word ;
 810          *                      ...
 811          *              }-
 812          *
 813          *      To make the tables more manageable, subpages are allowed
 814          *      and will be parsed recursively if $recursive == true.
 815          *
 816          */
 817         function parseCachedTable( $code, $subpage = '', $recursive = true ) {
 818                 global $wgMessageCache;
 819                 static $parsed = array();
 820
 821                 if ( !is_object( $wgMessageCache ) ) {
 822                         return array();
 823                 }
 824
 825                 $key = 'Conversiontable/' . $code;
 826                 if ( $subpage ) {
 827                         $key .= '/' . $subpage;
 828                 }
 829                 if ( array_key_exists( $key, $parsed ) ) {
 830                         return array();
 831                 }
 832
 833                 if ( strpos( $code, '/' ) === false ) {
 834                         $txt = $wgMessageCache->get( 'Conversiontable', true, $code );
 835                         if ( $txt === false ) {
 836                                 # FIXME: this method doesn't seem to be expecting
 837                                 # this possible outcome...
 838                                 $txt = '&lt;Conversiontable&gt;';
 839                         }
 840                 } else {
 841                         $title = Title::makeTitleSafe( NS_MEDIAWIKI,
 842                                                                                    "Conversiontable/$code" );
 843                         if ( $title && $title->exists() ) {
 844                                 $article = new Article( $title );
 845                                 $txt = $article->getContents();
 846                         } else {
 847                                 $txt = '';
 848                         }
 849                 }
 850
 851                 // get all subpage links of the form
 852                 // [[MediaWiki:conversiontable/zh-xx/...|...]]
 853                 $linkhead = $this->mLangObj->getNsText( NS_MEDIAWIKI ) .
 854                         ':Conversiontable';
 855                 $subs = StringUtils::explode( '[[', $txt );
 856                 $sublinks = array();
 857                 foreach ( $subs as $sub ) {
 858                         $link = explode( ']]', $sub, 2 );
 859                         if ( count( $link ) != 2 ) {
 860                                 continue;
 861                         }
 862                         $b = explode( '|', $link[0], 2 );
 863                         $b = explode( '/', trim( $b[0] ), 3 );
 864                         if ( count( $b ) == 3 ) {
 865                                 $sublink = $b[2];
 866                         } else {
 867                                 $sublink = '';
 868                         }
 869
 870                         if ( $b[0] == $linkhead && $b[1] == $code ) {
 871                                 $sublinks[] = $sublink;
 872                         }
 873                 }
 874
 875                 // parse the mappings in this page
 876                 $blocks = StringUtils::explode( '-{', $txt );
 877                 $ret = array();
 878                 $first = true;
 879                 foreach ( $blocks as $block ) {
 880                         if ( $first ) {
 881                                 // Skip the part before the first -{
 882                                 $first = false;
 883                                 continue;
 884                         }
 885                         $mappings = explode( '}-', $block, 2 );
 886                         $stripped = str_replace( array( "'", '"', '*', '#' ), '',
 887                                                                          $mappings[0] );
 888                         $table = StringUtils::explode( ';', $stripped );
 889                         foreach ( $table as $t ) {
 890                                 $m = explode( '=>', $t, 3 );
 891                                 if ( count( $m ) != 2 )
 892                                         continue;
 893                                 // trim any trailling comments starting with '//'
 894                                 $tt = explode( '//', $m[1], 2 );
 895                                 $ret[trim( $m[0] )] = trim( $tt[0] );
 896                         }
 897                 }
 898                 $parsed[$key] = true;
 899
 900                 // recursively parse the subpages
 901                 if ( $recursive ) {
 902                         foreach ( $sublinks as $link ) {
 903                                 $s = $this->parseCachedTable( $code, $link, $recursive );
 904                                 $ret = array_merge( $ret, $s );
 905                         }
 906                 }
 907
 908                 if ( $this->mUcfirst ) {
 909                         foreach ( $ret as $k => $v ) {
 910                                 $ret[$this->mLangObj->ucfirst( $k )] = $this->mLangObj->ucfirst( $v );
 911                         }
 912                 }
 913                 return $ret;
 914         }
 915
 916         /**
 917          * Enclose a string with the "no conversion" tag. This is used by
 918          * various functions in the Parser.
 919          *
 920          * @param $text String: text to be tagged for no conversion
 921          * @param $noParse Unused (?)
 922          * @return String: the tagged text
 923          */
 924         public function markNoConversion( $text, $noParse = false ) {
 925                 # don't mark if already marked
 926                 if ( strpos( $text, '-{' ) || strpos( $text, '}-' ) ) {
 927                         return $text;
 928                 }
 929
 930                 $ret = "-{R|$text}-";
 931                 return $ret;
 932         }
 933
 934         /**
 935          * Convert the sorting key for category links. This should make different
 936          * keys that are variants of each other map to the same key.
 937          */
 938         function convertCategoryKey( $key ) {
 939                 return $key;
 940         }
 941
 942         /**
 943          * Hook to refresh the cache of conversion tables when
 944          * MediaWiki:conversiontable* is updated.
 945          * @private
 946          */
 947         function OnArticleSaveComplete( $article, $user, $text, $summary, $isminor,
 948                         $iswatch, $section, $flags, $revision ) {
 949                 $titleobj = $article->getTitle();
 950                 if ( $titleobj->getNamespace() == NS_MEDIAWIKI ) {
 951                         $title = $titleobj->getDBkey();
 952                         $t = explode( '/', $title, 3 );
 953                         $c = count( $t );
 954                         if ( $c > 1 && $t[0] == 'Conversiontable' ) {
 955                                 if ( $this->validateVariant( $t[1] ) ) {
 956                                         $this->reloadTables();
 957                                 }
 958                         }
 959                 }
 960                 return true;
 961         }
 962
 963         /**
 964          * Armour rendered math against conversion.
 965          * Wrap math into rawoutput -{R| math }- syntax.
 966          */
 967         public function armourMath( $text ) {
 968                 // we need to convert '-{' and '}-' to '-&#123;' and '&#125;-'
 969                 // to avoid a unwanted '}-' appeared after the math-image.
 970                 $text = strtr( $text, array( '-{' => '-&#123;', '}-' => '&#125;-' ) );
 971                 $ret = "-{R|$text}-";
 972                 return $ret;
 973         }
 974
 975         /**
 976          * Get the cached separator pattern for ConverterRule::parseRules()
 977          */
 978         function getVarSeparatorPattern() {
 979                 if ( is_null( $this->mVarSeparatorPattern ) ) {
 980                         // varsep_pattern for preg_split:
 981                         // text should be splited by ";" only if a valid variant
 982                         // name exist after the markup, for example:
 983                         //  -{zh-hans:<span style="font-size:120%;">xxx</span>;zh-hant:\
 984                         //      <span style="font-size:120%;">yyy</span>;}-
 985                         // we should split it as:
 986                         //  array(
 987                         //        [0] => 'zh-hans:<span style="font-size:120%;">xxx</span>'
 988                         //        [1] => 'zh-hant:<span style="font-size:120%;">yyy</span>'
 989                         //        [2] => ''
 990                         //       )
 991                         $pat = '/;\s*(?=';
 992                         foreach ( $this->mVariants as $variant ) {
 993                                 // zh-hans:xxx;zh-hant:yyy
 994                                 $pat .= $variant . '\s*:|';
 995                                 // xxx=>zh-hans:yyy; xxx=>zh-hant:zzz
 996                                 $pat .= '[^;]*?=>\s*' . $variant . '\s*:|';
 997                         }
 998                         $pat .= '\s*$)/';
 999                         $this->mVarSeparatorPattern = $pat;
1000                 }
1001                 return $this->mVarSeparatorPattern;
1002         }
1003 }
1004
1005 /**
1006  * Parser for rules of language conversion , parse rules in -{ }- tag.
1007  * @ingroup Language
1008  * @author fdcn <fdcn64@gmail.com>, PhiLiP <philip.npc@gmail.com>
1009  */
1010 class ConverterRule {
1011         var $mText; // original text in -{text}-
1012         var $mConverter; // LanguageConverter object
1013         var $mManualCodeError = '<strong class="error">code error!</strong>';
1014         var $mRuleDisplay = '';
1015         var $mRuleTitle = false;
1016         var $mRules = '';// string : the text of the rules
1017         var $mRulesAction = 'none';
1018         var $mFlags = array();
1019         var $mVariantFlags = array();
1020         var $mConvTable = array();
1021         var $mBidtable = array();// array of the translation in each variant
1022         var $mUnidtable = array();// array of the translation in each variant
1023
1024         /**
1025          * Constructor
1026          *
1027          * @param $text String: the text between -{ and }-
1028          * @param $converter LanguageConverter object
1029          */
1030         public function __construct( $text, $converter ) {
1031                 $this->mText = $text;
1032                 $this->mConverter = $converter;
1033         }
1034
1035         /**
1036          * Check if variants array in convert array.
1037          *
1038          * @param $variants Array or string: variant language code
1039          * @return String: translated text
1040          */
1041         public function getTextInBidtable( $variants ) {
1042                 $variants = (array)$variants;
1043                 if ( !$variants ) {
1044                         return false;
1045                 }
1046                 foreach ( $variants as $variant ) {
1047                         if ( isset( $this->mBidtable[$variant] ) ) {
1048                                 return $this->mBidtable[$variant];
1049                         }
1050                 }
1051                 return false;
1052         }
1053
1054         /**
1055          * Parse flags with syntax -{FLAG| ... }-
1056          * @private
1057          */
1058         function parseFlags() {
1059                 $text = $this->mText;
1060                 $flags = array();
1061                 $variantFlags = array();
1062
1063                 $sepPos = strpos( $text, '|' );
1064                 if ( $sepPos !== false ) {
1065                         $validFlags = $this->mConverter->mFlags;
1066                         $f = StringUtils::explode( ';', substr( $text, 0, $sepPos ) );
1067                         foreach ( $f as $ff ) {
1068                                 $ff = trim( $ff );
1069                                 if ( isset( $validFlags[$ff] ) ) {
1070                                         $flags[$validFlags[$ff]] = true;
1071                                 }
1072                         }
1073                         $text = strval( substr( $text, $sepPos + 1 ) );
1074                 }
1075
1076                 if ( !$flags ) {
1077                         $flags['S'] = true;
1078                 } elseif ( isset( $flags['R'] ) ) {
1079                         $flags = array( 'R' => true );// remove other flags
1080                 } elseif ( isset( $flags['N'] ) ) {
1081                         $flags = array( 'N' => true );// remove other flags
1082                 } elseif ( isset( $flags['-'] ) ) {
1083                         $flags = array( '-' => true );// remove other flags
1084                 } elseif ( count( $flags ) == 1 && isset( $flags['T'] ) ) {
1085                         $flags['H'] = true;
1086                 } elseif ( isset( $flags['H'] ) ) {
1087                         // replace A flag, and remove other flags except T
1088                         $temp = array( '+' => true, 'H' => true );
1089                         if ( isset( $flags['T'] ) ) {
1090                                 $temp['T'] = true;
1091                         }
1092                         if ( isset( $flags['D'] ) ) {
1093                                 $temp['D'] = true;
1094                         }
1095                         $flags = $temp;
1096                 } else {
1097                         if ( isset( $flags['A'] ) ) {
1098                                 $flags['+'] = true;
1099                                 $flags['S'] = true;
1100                         }
1101                         if ( isset( $flags['D'] ) ) {
1102                                 unset( $flags['S'] );
1103                         }
1104                         // try to find flags like "zh-hans", "zh-hant"
1105                         // allow syntaxes like "-{zh-hans;zh-hant|XXXX}-"
1106                         $variantFlags = array_intersect( array_keys( $flags ), $this->mConverter->mVariants );
1107                         if ( $variantFlags ) {
1108                                 $variantFlags = array_flip( $variantFlags );
1109                                 $flags = array();
1110                         }
1111                 }
1112                 $this->mVariantFlags = $variantFlags;
1113                 $this->mRules = $text;
1114                 $this->mFlags = $flags;
1115         }
1116
1117         /**
1118          * Generate conversion table.
1119          * @private
1120          */
1121         function parseRules() {
1122                 $rules = $this->mRules;
1123                 $bidtable = array();
1124                 $unidtable = array();
1125                 $variants = $this->mConverter->mVariants;
1126                 $varsep_pattern = $this->mConverter->getVarSeparatorPattern();
1127
1128                 $choice = preg_split( $varsep_pattern, $rules );
1129
1130                 foreach ( $choice as $c ) {
1131                         $v  = explode( ':', $c, 2 );
1132                         if ( count( $v ) != 2 ) {
1133                                 // syntax error, skip
1134                                 continue;
1135                         }
1136                         $to = trim( $v[1] );
1137                         $v  = trim( $v[0] );
1138                         $u  = explode( '=>', $v, 2 );
1139                         // if $to is empty, strtr() could return a wrong result
1140                         if ( count( $u ) == 1 && $to && in_array( $v, $variants ) ) {
1141                                 $bidtable[$v] = $to;
1142                         } elseif ( count( $u ) == 2 ) {
1143                                 $from = trim( $u[0] );
1144                                 $v      = trim( $u[1] );
1145                                 if ( array_key_exists( $v, $unidtable )
1146                                          && !is_array( $unidtable[$v] )
1147                                          && $to
1148                                          && in_array( $v, $variants ) ) {
1149                                         $unidtable[$v] = array( $from => $to );
1150                                 } elseif ( $to && in_array( $v, $variants ) ) {
1151                                         $unidtable[$v][$from] = $to;
1152                                 }
1153                         }
1154                         // syntax error, pass
1155                         if ( !isset( $this->mConverter->mVariantNames[$v] ) ) {
1156                                 $bidtable = array();
1157                                 $unidtable = array();
1158                                 break;
1159                         }
1160                 }
1161                 $this->mBidtable = $bidtable;
1162                 $this->mUnidtable = $unidtable;
1163         }
1164
1165         /**
1166          * @private
1167          */
1168         function getRulesDesc() {
1169                 $codesep = $this->mConverter->mDescCodeSep;
1170                 $varsep = $this->mConverter->mDescVarSep;
1171                 $text = '';
1172                 foreach ( $this->mBidtable as $k => $v ) {
1173                         $text .= $this->mConverter->mVariantNames[$k] . "$codesep$v$varsep";
1174                 }
1175                 foreach ( $this->mUnidtable as $k => $a ) {
1176                         foreach ( $a as $from => $to ) {
1177                                 $text .= $from . '⇒' . $this->mConverter->mVariantNames[$k] .
1178                                         "$codesep$to$varsep";
1179                         }
1180                 }
1181                 return $text;
1182         }
1183
1184         /**
1185          * Parse rules conversion.
1186          * @private
1187          */
1188         function getRuleConvertedStr( $variant ) {
1189                 $bidtable = $this->mBidtable;
1190                 $unidtable = $this->mUnidtable;
1191
1192                 if ( count( $bidtable ) + count( $unidtable ) == 0 ) {
1193                         return $this->mRules;
1194                 } else {
1195                         // display current variant in bidirectional array
1196                         $disp = $this->getTextInBidtable( $variant );
1197                         // or display current variant in fallbacks
1198                         if ( !$disp ) {
1199                                 $disp = $this->getTextInBidtable(
1200                                                 $this->mConverter->getVariantFallbacks( $variant ) );
1201                         }
1202                         // or display current variant in unidirectional array
1203                         if ( !$disp && array_key_exists( $variant, $unidtable ) ) {
1204                                 $disp = array_values( $unidtable[$variant] );
1205                                 $disp = $disp[0];
1206                         }
1207                         // or display frist text under disable manual convert
1208                         if ( !$disp
1209                                  && $this->mConverter->mManualLevel[$variant] == 'disable' ) {
1210                                 if ( count( $bidtable ) > 0 ) {
1211                                         $disp = array_values( $bidtable );
1212                                         $disp = $disp[0];
1213                                 } else {
1214                                         $disp = array_values( $unidtable );
1215                                         $disp = array_values( $disp[0] );
1216                                         $disp = $disp[0];
1217                                 }
1218                         }
1219                         return $disp;
1220                 }
1221         }
1222
1223         /**
1224          * Generate conversion table for all text.
1225          * @private
1226          */
1227         function generateConvTable() {
1228                 // Special case optimisation
1229                 if ( !$this->mBidtable && !$this->mUnidtable ) {
1230                         $this->mConvTable = array();
1231                         return;
1232                 }
1233
1234                 $bidtable = $this->mBidtable;
1235                 $unidtable = $this->mUnidtable;
1236                 $manLevel = $this->mConverter->mManualLevel;
1237
1238                 $vmarked = array();
1239                 foreach ( $this->mConverter->mVariants as $v ) {
1240                         /* for bidirectional array
1241                                 fill in the missing variants, if any,
1242                                 with fallbacks */
1243                         if ( !isset( $bidtable[$v] ) ) {
1244                                 $variantFallbacks =
1245                                         $this->mConverter->getVariantFallbacks( $v );
1246                                 $vf = $this->getTextInBidtable( $variantFallbacks );
1247                                 if ( $vf ) {
1248                                         $bidtable[$v] = $vf;
1249                                 }
1250                         }
1251
1252                         if ( isset( $bidtable[$v] ) ) {
1253                                 foreach ( $vmarked as $vo ) {
1254                                         // use syntax: -{A|zh:WordZh;zh-tw:WordTw}-
1255                                         // or -{H|zh:WordZh;zh-tw:WordTw}-
1256                                         // or -{-|zh:WordZh;zh-tw:WordTw}-
1257                                         // to introduce a custom mapping between
1258                                         // words WordZh and WordTw in the whole text
1259                                         if ( $manLevel[$v] == 'bidirectional' ) {
1260                                                 $this->mConvTable[$v][$bidtable[$vo]] = $bidtable[$v];
1261                                         }
1262                                         if ( $manLevel[$vo] == 'bidirectional' ) {
1263                                                 $this->mConvTable[$vo][$bidtable[$v]] = $bidtable[$vo];
1264                                         }
1265                                 }
1266                                 $vmarked[] = $v;
1267                         }
1268                         /*for unidirectional array fill to convert tables */
1269                         if ( ( $manLevel[$v] == 'bidirectional' || $manLevel[$v] == 'unidirectional' )
1270                                 && isset( $unidtable[$v] ) )
1271                         {
1272                                 if ( isset( $this->mConvTable[$v] ) ) {
1273                                         $this->mConvTable[$v] = array_merge( $this->mConvTable[$v], $unidtable[$v] );
1274                                 } else {
1275                                         $this->mConvTable[$v] = $unidtable[$v];
1276                                 }
1277                         }
1278                 }
1279         }
1280
1281         /**
1282          * Parse rules and flags.
1283          * @public
1284          */
1285         function parse( $variant = NULL ) {
1286                 if ( !$variant ) {
1287                         $variant = $this->mConverter->getPreferredVariant();
1288                 }
1289
1290                 $this->parseFlags();
1291                 $flags = $this->mFlags;
1292
1293                 // convert to specified variant
1294                 // syntax: -{zh-hans;zh-hant[;...]|<text to convert>}-
1295                 if ( $this->mVariantFlags ) {
1296                         // check if current variant in flags
1297                         if ( isset( $this->mVariantFlags[$variant] ) ) {
1298                                 // then convert <text to convert> to current language
1299                                 $this->mRules = $this->mConverter->autoConvert( $this->mRules,
1300                                                                                                                                 $variant );
1301                         } else { // if current variant no in flags,
1302                                    // then we check its fallback variants.
1303                                 $variantFallbacks =
1304                                         $this->mConverter->getVariantFallbacks( $variant );
1305                                 foreach ( $variantFallbacks as $variantFallback ) {
1306                                         // if current variant's fallback exist in flags
1307                                         if ( isset( $this->mVariantFlags[$variantFallback] ) ) {
1308                                                 // then convert <text to convert> to fallback language
1309                                                 $this->mRules =
1310                                                         $this->mConverter->autoConvert( $this->mRules,
1311                                                                                                                         $variantFallback );
1312                                                 break;
1313                                         }
1314                                 }
1315                         }
1316                         $this->mFlags = $flags = array( 'R' => true );
1317                 }
1318
1319                 if ( !isset( $flags['R'] ) && !isset( $flags['N'] ) ) {
1320                         // decode => HTML entities modified by Sanitizer::removeHTMLtags
1321                         $this->mRules = str_replace( '=&gt;', '=>', $this->mRules );
1322                         $this->parseRules();
1323                 }
1324                 $rules = $this->mRules;
1325
1326                 if ( !$this->mBidtable && !$this->mUnidtable ) {
1327                         if ( isset( $flags['+'] ) || isset( $flags['-'] ) ) {
1328                                 // fill all variants if text in -{A/H/-|text} without rules
1329                                 foreach ( $this->mConverter->mVariants as $v ) {
1330                                         $this->mBidtable[$v] = $rules;
1331                                 }
1332                         } elseif ( !isset( $flags['N'] ) && !isset( $flags['T'] ) ) {
1333                                 $this->mFlags = $flags = array( 'R' => true );
1334                         }
1335                 }
1336
1337                 $this->mRuleDisplay = false;
1338                 foreach ( $flags as $flag => $unused ) {
1339                         switch ( $flag ) {
1340                                 case 'R':
1341                                         // if we don't do content convert, still strip the -{}- tags
1342                                         $this->mRuleDisplay = $rules;
1343                                         break;
1344                                 case 'N':
1345                                         // process N flag: output current variant name
1346                                         $ruleVar = trim( $rules );
1347                                         if ( isset( $this->mConverter->mVariantNames[$ruleVar] ) ) {
1348                                                 $this->mRuleDisplay = $this->mConverter->mVariantNames[$ruleVar];
1349                                         } else {
1350                                                 $this->mRuleDisplay = '';
1351                                         }
1352                                         break;
1353                                 case 'D':
1354                                         // process D flag: output rules description
1355                                         $this->mRuleDisplay = $this->getRulesDesc();
1356                                         break;
1357                                 case 'H':
1358                                         // process H,- flag or T only: output nothing
1359                                         $this->mRuleDisplay = '';
1360                                         break;
1361                                 case '-':
1362                                         $this->mRulesAction = 'remove';
1363                                         $this->mRuleDisplay = '';
1364                                         break;
1365                                 case '+':
1366                                         $this->mRulesAction = 'add';
1367                                         $this->mRuleDisplay = '';
1368                                         break;
1369                                 case 'S':
1370                                         $this->mRuleDisplay = $this->getRuleConvertedStr( $variant );
1371                                         break;
1372                                 case 'T':
1373                                         $this->mRuleTitle = $this->getRuleConvertedStr( $variant );
1374                                         $this->mRuleDisplay = '';
1375                                         break;
1376                                 default:
1377                                         // ignore unknown flags (but see error case below)
1378                         }
1379                 }
1380                 if ( $this->mRuleDisplay === false ) {
1381                         $this->mRuleDisplay = $this->mManualCodeError;
1382                 }
1383
1384                 $this->generateConvTable();
1385         }
1386
1387         /**
1388          * @public
1389          */
1390         function hasRules() {
1391                 // TODO:
1392         }
1393
1394         /**
1395          * Get display text on markup -{...}-
1396          * @public
1397          */
1398         function getDisplay() {
1399                 return $this->mRuleDisplay;
1400         }
1401
1402         /**
1403          * Get converted title.
1404          * @public
1405          */
1406         function getTitle() {
1407                 return $this->mRuleTitle;
1408         }
1409
1410         /**
1411          * Return how deal with conversion rules.
1412          * @public
1413          */
1414         function getRulesAction() {
1415                 return $this->mRulesAction;
1416         }
1417
1418         /**
1419          * Get conversion table. ( bidirectional and unidirectional
1420          * conversion table )
1421          * @public
1422          */
1423         function getConvTable() {
1424                 return $this->mConvTable;
1425         }
1426
1427         /**
1428          * Get conversion rules string.
1429          * @public
1430          */
1431         function getRules() {
1432                 return $this->mRules;
1433         }
1434
1435         /**
1436          * Get conversion flags.
1437          * @public
1438          */
1439         function getFlags() {
1440                 return $this->mFlags;
1441         }
1442 }