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