languages/LanguageConverter.php

   1 <?php
   2
   3 /**
   4  * Contains the LanguageConverter class and ConverterRule class
   5  * @ingroup Language
   6  *
   7  * @license http://www.gnu.org/copyleft/gpl.html GNU General Public License
   8  * @file
   9  */
  10
  11 /**
  12  * Base class for language conversion.
  13  * @ingroup Language
  14  *
  15  * @author Zhengzhu Feng <zhengzhu@gmail.com>
  16  * @maintainers fdcn <fdcn64@gmail.com>, shinjiman <shinjiman@gmail.com>, PhiLiP <philip.npc@gmail.com>
  17  */
  18 class LanguageConverter {
  19         var $mMainLanguageCode;
  20         var $mVariants, $mVariantFallbacks, $mVariantNames;
  21         var $mTablesLoaded = false;
  22         var $mTables;
  23         // 'bidirectional' 'unidirectional' 'disable' for each variant
  24         var $mManualLevel;
  25         var $mCacheKey;
  26         var $mLangObj;
  27         var $mFlags;
  28         var $mDescCodeSep = ':', $mDescVarSep = ';';
  29         var $mUcfirst = false;
  30         var $mConvRuleTitle = false;
  31         var $mURLVariant;
  32         var $mUserVariant;
  33         var $mHeaderVariant;
  34         var $mMaxDepth = 10;
  35         var $mVarSeparatorPattern;
  36
  37         const CACHE_VERSION_KEY = 'VERSION 6';
  38
  39         /**
  40          * Constructor
  41          *
  42          * @param $langobj The Language Object
  43          * @param $maincode String: the main language code of this language
  44          * @param $variants Array: the supported variants of this language
  45          * @param $variantfallbacks Array: the fallback language of each variant
  46          * @param $flags Array: defining the custom strings that maps to the flags
  47          * @param $manualLevel Array: limit for supported variants
  48          */
  49         public function __construct( $langobj, $maincode,
  50                                                                 $variants = array(),
  51                                                                 $variantfallbacks = array(),
  52                                                                 $flags = array(),
  53                                                                 $manualLevel = array() ) {
  54                 global $wgDisabledVariants, $wgLanguageNames;
  55                 $this->mLangObj = $langobj;
  56                 $this->mMainLanguageCode = $maincode;
  57                 $this->mVariants = array_diff( $variants, $wgDisabledVariants );
  58                 $this->mVariantFallbacks = $variantfallbacks;
  59                 $this->mVariantNames = $wgLanguageNames;
  60                 $this->mCacheKey = wfMemcKey( 'conversiontables', $maincode );
  61                 $defaultflags = array(
  62                         // 'S' show converted text
  63                         // '+' add rules for alltext
  64                         // 'E' the gave flags is error
  65                         // these flags above are reserved for program
  66                         'A' => 'A',       // add rule for convert code (all text convert)
  67                         'T' => 'T',       // title convert
  68                         'R' => 'R',       // raw content
  69                         'D' => 'D',       // convert description (subclass implement)
  70                         '-' => '-',       // remove convert (not implement)
  71                         'H' => 'H',       // add rule for convert code
  72                                                   // (but no display in placed code )
  73                         'N' => 'N'        // current variant name
  74                 );
  75                 $this->mFlags = array_merge( $defaultflags, $flags );
  76                 foreach ( $this->mVariants as $v ) {
  77                         if ( array_key_exists( $v, $manualLevel ) ) {
  78                                 $this->mManualLevel[$v] = $manualLevel[$v];
  79                         } else {
  80                                 $this->mManualLevel[$v] = 'bidirectional';
  81                         }
  82                         $this->mFlags[$v] = $v;
  83                 }
  84         }
  85
  86         /**
  87          * Get all valid variants.
  88          * Call this instead of using $this->mVariants directly.
  89          *
  90          * @return Array: contains all valid variants
  91          */
  92         public function getVariants() {
  93                 return $this->mVariants;
  94         }
  95
  96         /**
  97          * In case some variant is not defined in the markup, we need
  98          * to have some fallback. For example, in zh, normally people
  99          * will define zh-hans and zh-hant, but less so for zh-sg or zh-hk.
 100          * when zh-sg is preferred but not defined, we will pick zh-hans
 101          * in this case. Right now this is only used by zh.
 102          *
 103          * @param $variant String: the language code of the variant
 104          * @return String: The code of the fallback language or the
 105          *                               main code if there is no fallback
 106          */
 107         public function getVariantFallbacks( $variant ) {
 108                 if ( isset( $this->mVariantFallbacks[$variant] ) ) {
 109                         return $this->mVariantFallbacks[$variant];
 110                 }
 111                 return $this->mMainLanguageCode;
 112         }
 113
 114         /**
 115          * Get the title produced by the conversion rule.
 116          * @return String: The converted title text
 117          */
 118         public function getConvRuleTitle() {
 119                 return $this->mConvRuleTitle;
 120         }
 121
 122         /**
 123          * Get preferred language variant.
 124          * @param $fromUser Boolean: get it from $wgUser's preferences
 125          * @param $fromHeader Boolean: get it from Accept-Language
 126          * @return String: the preferred language code
 127          */
 128         public function getPreferredVariant( $fromUser = true, $fromHeader = false ) {
 129                 global $wgDefaultLanguageVariant;
 130
 131                 $req = $this->getURLVariant();
 132
 133                 if ( $fromUser && !$req ) {
 134                         $req = $this->getUserVariant();
 135                 }
 136
 137                 if ( $fromHeader && !$req ) {
 138                         $req = $this->getHeaderVariant();
 139                 }
 140
 141                 if ( $wgDefaultLanguageVariant && !$req ) {
 142                         $req = $this->validateVariant( $wgDefaultLanguageVariant );
 143                 }
 144
 145                 // This function, unlike the other get*Variant functions, is
 146                 // not memoized (i.e. there return value is not cached) since
 147                 // new information might appear during processing after this
 148                 // is first called.
 149                 if ( $req ) {
 150                         return $req;
 151                 }
 152                 return $this->mMainLanguageCode;
 153         }
 154
 155         /**
 156          * Validate the variant
 157          * @param $variant String: the variant to validate
 158          * @return Mixed: returns the variant if it is valid, null otherwise
 159          */
 160         protected function validateVariant( $variant = null ) {
 161                 if ( $variant !== null &&
 162                          in_array( $variant, $this->mVariants ) ) {
 163                         return $variant;
 164                 }
 165                 return null;
 166         }
 167
 168         /**
 169          * Get the variant specified in the URL
 170          *
 171          * @return Mixed: variant if one found, false otherwise.
 172          */
 173         protected function getURLVariant() {
 174                 global $wgRequest;
 175                 $ret = null;
 176
 177                 if ( $this->mURLVariant ) {
 178                         return $this->mURLVariant;
 179                 }
 180
 181                 // see if the preference is set in the request
 182                 $ret = $wgRequest->getText( 'variant' );
 183
 184                 if ( !$ret ) {
 185                         $ret = $wgRequest->getVal( 'uselang' );
 186                 }
 187
 188                 return $this->mURLVariant = $this->validateVariant( $ret );
 189         }
 190
 191         /**
 192          * Determine if the user has a variant set.
 193          *
 194          * @return Mixed: variant if one found, false otherwise.
 195          */
 196         protected function getUserVariant() {
 197                 global $wgUser;
 198                 $ret = null;
 199
 200                 // memoizing this function wreaks havoc on parserTest.php
 201                 /* if ( $this->mUserVariant ) { */
 202                 /*      return $this->mUserVariant; */
 203                 /* } */
 204
 205                 // get language variant preference from logged in users
 206                 // Don't call this on stub objects because that causes infinite
 207                 // recursion during initialisation
 208                 if ( $wgUser->isLoggedIn() )  {
 209                         $ret = $wgUser->getOption( 'variant' );
 210                 }
 211                 else {
 212                         // figure out user lang without constructing wgLang to avoid
 213                         // infinite recursion
 214                         $ret = $wgUser->getOption( 'language' );
 215                 }
 216
 217                 return $this->mUserVariant = $this->validateVariant( $ret );
 218         }
 219
 220
 221         /**
 222          * Determine the language variant from the Accept-Language header.
 223          *
 224          * @return Mixed: variant if one found, false otherwise.
 225          */
 226         protected function getHeaderVariant() {
 227                 global $wgRequest;
 228                 $ret = null;
 229
 230                 if ( $this->mHeaderVariant ) {
 231                         return $this->mHeaderVariant;
 232                 }
 233
 234                 // see if some supported language variant is set in the
 235                 // http header.
 236                 $acceptLanguage = $wgRequest->getHeader( 'Accept-Language' );
 237                 if ( !$acceptLanguage ) {
 238                         return null;
 239                 }
 240
 241                 // explode by comma
 242                 $result = StringUtils::explode( ',', strtolower( $acceptLanguage ) );
 243                 $languages = array();
 244
 245                 foreach ( $result as $elem ) {
 246                         // if $elem likes 'zh-cn;q=0.9'
 247                         if ( ( $posi = strpos( $elem, ';' ) ) !== false ) {
 248                                 // get the real language code likes 'zh-cn'
 249                                 $languages[] = substr( $elem, 0, $posi );
 250                         } else {
 251                                 $languages[] = $elem;
 252                         }
 253                 }
 254
 255                 $fallback_languages = array();
 256                 foreach ( $languages as $language ) {
 257                         // strip whitespace
 258                         $language = trim( $language );
 259                         $this->mHeaderVariant = $this->validateVariant( $language );
 260                         if ( $this->mHeaderVariant ) {
 261                                 break;
 262                         }
 263
 264                         // To see if there are fallbacks of current language.
 265                         // We record these fallback variants, and process
 266                         // them later.
 267                         $fallbacks = $this->getVariantFallbacks( $language );
 268                         if ( is_string( $fallbacks ) ) {
 269                                 $fallback_languages[] = $fallbacks;
 270                         } elseif ( is_array( $fallbacks ) ) {
 271                                 $fallback_languages =
 272                                         array_merge( $fallback_languages,
 273                                                                  $fallbacks );
 274                         }
 275                 }
 276
 277                 if ( !$this->mHeaderVariant ) {
 278                         // process fallback languages now
 279                         $fallback_languages = array_unique( $fallback_languages );
 280                         foreach ( $fallback_languages as $language ) {
 281                                 $this->mHeaderVariant = $this->validateVariant( $language );
 282                                 if ( $this->mHeaderVariant ) {
 283                                         break;
 284                                 }
 285                         }
 286                 }
 287
 288                 return $this->mHeaderVariant;
 289         }
 290
 291         /**
 292          * Caption convert, base on preg_replace_callback.
 293          *
 294          * To convert text in "title" or "alt", like '<img alt="text" ... '
 295          * or '<span title="text" ... '
 296          *
 297          * @return String like ' alt="yyyy"' or ' title="yyyy"'
 298          */
 299         protected function captionConvert( $matches ) {
 300                 $toVariant = $this->getPreferredVariant();
 301                 $title = $matches[1];
 302                 $text  = $matches[2];
 303                 // we convert captions except URL
 304                 if ( !strpos( $text, '://' ) ) {
 305                         $text = $this->translate( $text, $toVariant );
 306                 }
 307                 return " $title=\"$text\"";
 308         }
 309
 310         /**
 311          * Dictionary-based conversion.
 312          * This function would not parse the conversion rules.
 313          * If you want to parse rules, try to use convert() or
 314          * convertTo().
 315          *
 316          * @param $text String: the text to be converted
 317          * @param $toVariant String: the target language code
 318          * @return String: the converted text
 319          */
 320         public function autoConvert( $text, $toVariant = false ) {
 321                 wfProfileIn( __METHOD__ );
 322
 323                 $this->loadTables();
 324
 325                 if ( !$toVariant ) {
 326                         $toVariant = $this->getPreferredVariant();
 327                         if ( !$toVariant ) {
 328                                 return $text;
 329                         }
 330                 }
 331
 332                 /* we convert everything except:
 333                    1. html markups (anything between < and >)
 334                    2. html entities
 335                    3. place holders created by the parser
 336                 */
 337                 global $wgParser;
 338                 if ( isset( $wgParser ) && $wgParser->UniqPrefix() != '' ) {
 339                         $marker = '|' . $wgParser->UniqPrefix() . '[\-a-zA-Z0-9]+';
 340                 } else {
 341                         $marker = '';
 342                 }
 343
 344                 // this one is needed when the text is inside an html markup
 345                 $htmlfix = '|<[^>]+$|^[^<>]*>';
 346
 347                 // disable convert to variants between <code></code> tags
 348                 $codefix = '<code>.+?<\/code>|';
 349                 // disable convertsion of <script type="text/javascript"> ... </script>
 350                 $scriptfix = '<script.*?>.*?<\/script>|';
 351                 // disable conversion of <pre xxxx> ... </pre>
 352                 $prefix = '<pre.*?>.*?<\/pre>|';
 353
 354                 $reg = '/' . $codefix . $scriptfix . $prefix .
 355                         '<[^>]+>|&[a-zA-Z#][a-z0-9]+;' . $marker . $htmlfix . '/s';
 356
 357                 $matches = preg_split( $reg, $text, - 1, PREG_SPLIT_OFFSET_CAPTURE );
 358
 359                 $m = array_shift( $matches );
 360
 361                 $ret = $this->translate( $m[0], $toVariant );
 362                 $mstart = $m[1] + strlen( $m[0] );
 363
 364                 // enable convertsion of '<img alt="xxxx" ... '
 365                 // or '<span title="xxxx" ... '
 366                 $captionpattern  = '/\s(title|alt)\s*=\s*"([\s\S]*?)"/';
 367
 368                 $trtext = '';
 369                 $trtextmark = "\0";
 370                 $notrtext = array();
 371                 foreach ( $matches as $m ) {
 372                         $mark = substr( $text, $mstart, $m[1] - $mstart );
 373                         $mark = preg_replace_callback( $captionpattern,
 374                                                                                    array( &$this, 'captionConvert' ),
 375                                                                                    $mark );
 376                         // Let's convert the trtext only once,
 377                         // it would give us more performance improvement
 378                         $notrtext[] = $mark;
 379                         $trtext .= $m[0] . $trtextmark;
 380                         $mstart = $m[1] + strlen( $m[0] );
 381                 }
 382                 $notrtext[] = '';
 383                 $trtext = $this->translate( $trtext, $toVariant );
 384                 $trtext = StringUtils::explode( $trtextmark, $trtext );
 385                 foreach ( $trtext as $t ) {
 386                         $ret .= array_shift( $notrtext );
 387                         $ret .= $t;
 388                 }
 389                 wfProfileOut( __METHOD__ );
 390                 return $ret;
 391         }
 392
 393         /**
 394          * Translate a string to a variant.
 395          * Doesn't parse rules or do any of that other stuff, for that use
 396          * convert() or convertTo().
 397          *
 398          * @param $text String: text to convert
 399          * @param $variant String: variant language code
 400          * @return String: translated text
 401          */
 402         protected function translate( $text, $variant ) {
 403                 wfProfileIn( __METHOD__ );
 404                 // If $text is empty or only includes spaces, do nothing
 405                 // Otherwise translate it
 406                 if ( trim( $text ) ) {
 407                         $this->loadTables();
 408                         $text = $this->mTables[$variant]->replace( $text );
 409                 }
 410                 wfProfileOut( __METHOD__ );
 411                 return $text;
 412         }
 413
 414         /**
 415          * Call translate() to convert text to all valid variants.
 416          *
 417          * @param $text String: the text to be converted
 418          * @return Array: variant => converted text
 419          */
 420         public function autoConvertToAllVariants( $text ) {
 421                 wfProfileIn( __METHOD__ );
 422                 $this->loadTables();
 423
 424                 $ret = array();
 425                 foreach ( $this->mVariants as $variant ) {
 426                         $ret[$variant] = $this->translate( $text, $variant );
 427                 }
 428
 429                 wfProfileOut( __METHOD__ );
 430                 return $ret;
 431         }
 432
 433         /**
 434          * Convert link text to all valid variants.
 435          * In the first, this function only convert text outside the
 436          * "-{" "}-" markups. Since the "{" and "}" are not allowed in
 437          * titles, the text will get all converted always.
 438          * So I removed this feature and deprecated the function.
 439          *
 440          * @param $text String: the text to be converted
 441          * @return Array: variant => converted text
 442          * @deprecated Use autoConvertToAllVariants() instead
 443          */
 444         public function convertLinkToAllVariants( $text ) {
 445                 return $this->autoConvertToAllVariants( $text );
 446         }
 447
 448         /**
 449          * Apply manual conversion rules.
 450          *
 451          * @param $convRule Object: Object of ConverterRule
 452          */
 453         protected function applyManualConv( $convRule ) {
 454                 // Use syntax -{T|zh-cn:TitleCN; zh-tw:TitleTw}- to custom
 455                 // title conversion.
 456                 // Bug 24072: $mConvRuleTitle was overwritten by other manual
 457                 // rule(s) not for title, this breaks the title conversion.
 458                 $newConvRuleTitle = $convRule->getTitle();
 459                 if( $newConvRuleTitle ) {
 460                         // So I add an empty check for getTitle()
 461                         $this->mConvRuleTitle = $newConvRuleTitle;
 462                 }
 463
 464                 // merge/remove manual conversion rules to/from global table
 465                 $convTable = $convRule->getConvTable();
 466                 $action = $convRule->getRulesAction();
 467                 foreach ( $convTable as $variant => $pair ) {
 468                         if ( !$this->validateVariant( $variant ) ) {
 469                                 continue;
 470                         }
 471
 472                         if ( $action == 'add' ) {
 473                                 foreach ( $pair as $from => $to ) {
 474                                         // to ensure that $from and $to not be left blank
 475                                         // so $this->translate() could always return a string
 476                                         if ( $from || $to ) {
 477                                                 // more efficient than array_merge(), about 2.5 times.
 478                                                 $this->mTables[$variant]->setPair( $from, $to );
 479                                         }
 480                                 }
 481                         } elseif ( $action == 'remove' ) {
 482                                 $this->mTables[$variant]->removeArray( $pair );
 483                         }
 484                 }
 485         }
 486
 487         /**
 488          * Auto convert a Title object to a readable string in the
 489          * preferred variant.
 490          *
 491          *@param $title Object: a object of Title
 492          *@return String: converted title text
 493          */
 494         public function convertTitle( $title ) {
 495                 $variant = $this->getPreferredVariant();
 496                 $index = $title->getNamespace();
 497                 if ( $index === NS_MAIN ) {
 498                         $text = '';
 499                 } else {
 500                         // first let's check if a message has given us a converted name
 501                         $nsConvKey = 'conversion-ns' . $index;
 502                         if ( !wfEmptyMsg( $nsConvKey ) ) {
 503                                 $text = wfMsgForContentNoTrans( $nsConvKey );
 504                         } else {
 505                                 // the message does not exist, try retrieve it from the current
 506                                 // variant's namespace names.
 507                                 $langObj = $this->mLangObj->factory( $variant );
 508                                 $text = $langObj->getFormattedNsText( $index );
 509                         }
 510                         $text .= ':';
 511                 }
 512                 $text .= $title->getText();
 513                 $text = $this->translate( $text, $variant );
 514                 return $text;
 515         }
 516
 517         /**
 518          * Convert text to different variants of a language. The automatic
 519          * conversion is done in autoConvert(). Here we parse the text
 520          * marked with -{}-, which specifies special conversions of the
 521          * text that can not be accomplished in autoConvert().
 522          *
 523          * Syntax of the markup:
 524          * -{code1:text1;code2:text2;...}-  or
 525          * -{flags|code1:text1;code2:text2;...}-  or
 526          * -{text}- in which case no conversion should take place for text
 527          *
 528          * @param $text String: text to be converted
 529          * @return String: converted text
 530          */
 531         public function convert( $text ) {
 532                 $variant = $this->getPreferredVariant();
 533                 return $this->convertTo( $text, $variant );
 534         }
 535
 536         /**
 537          * Same as convert() except a extra parameter to custom variant.
 538          *
 539          * @param $text String: text to be converted
 540          * @param $variant String: the target variant code
 541          * @return String: converted text
 542          */
 543         public function convertTo( $text, $variant ) {
 544                 global $wgDisableLangConversion;
 545                 if ( $wgDisableLangConversion ) return $text;
 546                 return $this->recursiveConvertTopLevel( $text, $variant );
 547         }
 548
 549         /**
 550          * Recursively convert text on the outside. Allow to use nested
 551          * markups to custom rules.
 552          *
 553          * @param $text String: text to be converted
 554          * @param $variant String: the target variant code
 555          * @param $depth Integer: depth of recursion
 556          * @return String: converted text
 557          */
 558         protected function recursiveConvertTopLevel( $text, $variant, $depth = 0 ) {
 559                 $startPos = 0;
 560                 $out = '';
 561                 $length = strlen( $text );
 562                 while ( $startPos < $length ) {
 563                         $m = false;
 564                         $pos = strpos( $text, '-{', $startPos );
 565
 566                         if ( $pos === false ) {
 567                                 // No more markup, append final segment
 568                                 $out .= $this->autoConvert( substr( $text, $startPos ), $variant );
 569                                 $startPos = $length;
 570                                 return $out;
 571                         }
 572
 573                         // Markup found
 574                         // Append initial segment
 575                         $out .= $this->autoConvert( substr( $text, $startPos, $pos - $startPos ), $variant );
 576
 577                         // Advance position
 578                         $startPos = $pos;
 579
 580                         // Do recursive conversion
 581                         $out .= $this->recursiveConvertRule( $text, $variant, $startPos, $depth + 1 );
 582                 }
 583
 584                 return $out;
 585         }
 586
 587         /**
 588          * Recursively convert text on the inside.
 589          *
 590          * @param $text String: text to be converted
 591          * @param $variant String: the target variant code
 592          * @param $depth Integer: depth of recursion
 593          * @return String: converted text
 594          */
 595         protected function recursiveConvertRule( $text, $variant, &$startPos, $depth = 0 ) {
 596                 // Quick sanity check (no function calls)
 597                 if ( $text[$startPos] !== '-' || $text[$startPos + 1] !== '{' ) {
 598                         throw new MWException( __METHOD__.': invalid input string' );
 599                 }
 600
 601                 $startPos += 2;
 602                 $inner = '';
 603                 $warningDone = false;
 604                 $length = strlen( $text );
 605
 606                 while ( $startPos < $length ) {
 607                         $m = false;
 608                         preg_match( '/-\{|\}-/', $text, $m,  PREG_OFFSET_CAPTURE, $startPos );
 609                         if ( !$m ) {
 610                                 // Unclosed rule
 611                                 break;
 612                         }
 613
 614                         $token = $m[0][0];
 615                         $pos = $m[0][1];
 616
 617                         // Markup found
 618                         // Append initial segment
 619                         $inner .= substr( $text, $startPos, $pos - $startPos );
 620
 621                         // Advance position
 622                         $startPos = $pos;
 623
 624                         switch ( $token ) {
 625                                 case '-{':
 626                                         // Check max depth
 627                                         if ( $depth >= $this->mMaxDepth ) {
 628                                                 $inner .= '-{';
 629                                                 if ( !$warningDone ) {
 630                                                         $inner .= '<span class="error">' .
 631                                                                 wfMsgForContent( 'language-converter-depth-warning',
 632                                                                         $this->mMaxDepth ) .
 633                                                                 '</span>';
 634                                                         $warningDone = true;
 635                                                 }
 636                                                 $startPos += 2;
 637                                                 continue;
 638                                         }
 639                                         // Recursively parse another rule
 640                                         $inner .= $this->recursiveConvertRule( $text, $variant, $startPos, $depth + 1 );
 641                                         break;
 642                                 case '}-':
 643                                         // Apply the rule
 644                                         $startPos += 2;
 645                                         $rule = new ConverterRule( $inner, $this );
 646                                         $rule->parse( $variant );
 647                                         $this->applyManualConv( $rule );
 648                                         return $rule->getDisplay();
 649                                 default:
 650                                         throw new MWException( __METHOD__.': invalid regex match' );
 651                         }
 652                 }
 653
 654                 // Unclosed rule
 655                 if ( $startPos < $length ) {
 656                         $inner .= substr( $text, $startPos );
 657                 }
 658                 $startPos = $length;
 659                 return '-{' . $this->autoConvert( $inner, $variant );
 660         }
 661
 662         /**
 663          * If a language supports multiple variants, it is
 664          * possible that non-existing link in one variant
 665          * actually exists in another variant. This function
 666          * tries to find it. See e.g. LanguageZh.php
 667          *
 668          * @param $link String: the name of the link
 669          * @param $nt Mixed: the title object of the link
 670          * @param $ignoreOtherCond Boolean: to disable other conditions when
 671          *              we need to transclude a template or update a category's link
 672          * @return Null, the input parameters may be modified upon return
 673          */
 674         public function findVariantLink( &$link, &$nt, $ignoreOtherCond = false ) {
 675                 # If the article has already existed, there is no need to
 676                 # check it again, otherwise it may cause a fault.
 677                 if ( is_object( $nt ) && $nt->exists() ) {
 678                         return;
 679                 }
 680
 681                 global $wgDisableLangConversion, $wgDisableTitleConversion, $wgRequest,
 682                         $wgUser;
 683                 $isredir = $wgRequest->getText( 'redirect', 'yes' );
 684                 $action = $wgRequest->getText( 'action' );
 685                 $linkconvert = $wgRequest->getText( 'linkconvert', 'yes' );
 686                 $disableLinkConversion = $wgDisableLangConversion
 687                         || $wgDisableTitleConversion;
 688                 $linkBatch = new LinkBatch();
 689
 690                 $ns = NS_MAIN;
 691
 692                 if ( $disableLinkConversion ||
 693                          ( !$ignoreOtherCond &&
 694                            ( $isredir == 'no'
 695                                  || $action == 'edit'
 696                                  || $action == 'submit'
 697                                  || $linkconvert == 'no'
 698                                  || $wgUser->getOption( 'noconvertlink' ) == 1 ) ) ) {
 699                         return;
 700                 }
 701
 702                 if ( is_object( $nt ) ) {
 703                         $ns = $nt->getNamespace();
 704                 }
 705
 706                 $variants = $this->autoConvertToAllVariants( $link );
 707                 if ( !$variants ) { // give up
 708                         return;
 709                 }
 710
 711                 $titles = array();
 712
 713                 foreach ( $variants as $v ) {
 714                         if ( $v != $link ) {
 715                                 $varnt = Title::newFromText( $v, $ns );
 716                                 if ( !is_null( $varnt ) ) {
 717                                         $linkBatch->addObj( $varnt );
 718                                         $titles[] = $varnt;
 719                                 }
 720                         }
 721                 }
 722
 723                 // fetch all variants in single query
 724                 $linkBatch->execute();
 725
 726                 foreach ( $titles as $varnt ) {
 727                         if ( $varnt->getArticleID() > 0 ) {
 728                                 $nt = $varnt;
 729                                 $link = $varnt->getText();
 730                                 break;
 731                         }
 732                 }
 733         }
 734
 735         /**
 736          * Returns language specific hash options.
 737          */
 738         public function getExtraHashOptions() {
 739                 $variant = $this->getPreferredVariant();
 740                 return '!' . $variant ;
 741         }
 742
 743         /**
 744          * Load default conversion tables.
 745          * This method must be implemented in derived class.
 746          *
 747          * @private
 748          */
 749         function loadDefaultTables() {
 750                 $name = get_class( $this );
 751                 wfDie( "Must implement loadDefaultTables() method in class $name" );
 752         }
 753
 754         /**
 755          * Load conversion tables either from the cache or the disk.
 756          * @private
 757          */
 758         function loadTables( $fromcache = true ) {
 759                 if ( $this->mTablesLoaded ) {
 760                         return;
 761                 }
 762                 global $wgMemc;
 763                 wfProfileIn( __METHOD__ );
 764                 $this->mTablesLoaded = true;
 765                 $this->mTables = false;
 766                 if ( $fromcache ) {
 767                         wfProfileIn( __METHOD__ . '-cache' );
 768                         $this->mTables = $wgMemc->get( $this->mCacheKey );
 769                         wfProfileOut( __METHOD__ . '-cache' );
 770                 }
 771                 if ( !$this->mTables
 772                          || !array_key_exists( self::CACHE_VERSION_KEY, $this->mTables ) ) {
 773                         wfProfileIn( __METHOD__ . '-recache' );
 774                         // not in cache, or we need a fresh reload.
 775                         // we will first load the default tables
 776                         // then update them using things in MediaWiki:Zhconversiontable/*
 777                         $this->loadDefaultTables();
 778                         foreach ( $this->mVariants as $var ) {
 779                                 $cached = $this->parseCachedTable( $var );
 780                                 $this->mTables[$var]->mergeArray( $cached );
 781                         }
 782
 783                         $this->postLoadTables();
 784                         $this->mTables[self::CACHE_VERSION_KEY] = true;
 785
 786                         $wgMemc->set( $this->mCacheKey, $this->mTables, 43200 );
 787                         wfProfileOut( __METHOD__ . '-recache' );
 788                 }
 789                 wfProfileOut( __METHOD__ );
 790         }
 791
 792         /**
 793          * Hook for post processig after conversion tables are loaded.
 794          *
 795          */
 796         function postLoadTables() { }
 797
 798         /**
 799          * Reload the conversion tables.
 800          *
 801          * @private
 802          */
 803         function reloadTables() {
 804                 if ( $this->mTables ) {
 805                         unset( $this->mTables );
 806                 }
 807                 $this->mTablesLoaded = false;
 808                 $this->loadTables( false );
 809         }
 810
 811
 812         /**
 813          * Parse the conversion table stored in the cache.
 814          *
 815          * The tables should be in blocks of the following form:
 816          *              -{
 817          *                      word => word ;
 818          *                      word => word ;
 819          *                      ...
 820          *              }-
 821          *
 822          *      To make the tables more manageable, subpages are allowed
 823          *      and will be parsed recursively if $recursive == true.
 824          *
 825          */
 826         function parseCachedTable( $code, $subpage = '', $recursive = true ) {
 827                 global $wgMessageCache;
 828                 static $parsed = array();
 829
 830                 if ( !is_object( $wgMessageCache ) ) {
 831                         return array();
 832                 }
 833
 834                 $key = 'Conversiontable/' . $code;
 835                 if ( $subpage ) {
 836                         $key .= '/' . $subpage;
 837                 }
 838                 if ( array_key_exists( $key, $parsed ) ) {
 839                         return array();
 840                 }
 841
 842                 if ( strpos( $code, '/' ) === false ) {
 843                         $txt = $wgMessageCache->get( 'Conversiontable', true, $code );
 844                         if( $txt === false ){
 845                                 # FIXME: this method doesn't seem to be expecting
 846                                 # this possible outcome...
 847                                 $txt = '&lt;Conversiontable&gt;';
 848                         }
 849                 } else {
 850                         $title = Title::makeTitleSafe( NS_MEDIAWIKI,
 851                                                                                    "Conversiontable/$code" );
 852                         if ( $title && $title->exists() ) {
 853                                 $article = new Article( $title );
 854                                 $txt = $article->getContents();
 855                         } else {
 856                                 $txt = '';
 857                         }
 858                 }
 859
 860                 // get all subpage links of the form
 861                 // [[MediaWiki:conversiontable/zh-xx/...|...]]
 862                 $linkhead = $this->mLangObj->getNsText( NS_MEDIAWIKI ) .
 863                         ':Conversiontable';
 864                 $subs = StringUtils::explode( '[[', $txt );
 865                 $sublinks = array();
 866                 foreach ( $subs as $sub ) {
 867                         $link = explode( ']]', $sub, 2 );
 868                         if ( count( $link ) != 2 ) {
 869                                 continue;
 870                         }
 871                         $b = explode( '|', $link[0], 2 );
 872                         $b = explode( '/', trim( $b[0] ), 3 );
 873                         if ( count( $b ) == 3 ) {
 874                                 $sublink = $b[2];
 875                         } else {
 876                                 $sublink = '';
 877                         }
 878
 879                         if ( $b[0] == $linkhead && $b[1] == $code ) {
 880                                 $sublinks[] = $sublink;
 881                         }
 882                 }
 883
 884
 885                 // parse the mappings in this page
 886                 $blocks = StringUtils::explode( '-{', $txt );
 887                 $ret = array();
 888                 $first = true;
 889                 foreach ( $blocks as $block ) {
 890                         if ( $first ) {
 891                                 // Skip the part before the first -{
 892                                 $first = false;
 893                                 continue;
 894                         }
 895                         $mappings = explode( '}-', $block, 2 );
 896                         $stripped = str_replace( array( "'", '"', '*', '#' ), '',
 897                                                                          $mappings[0] );
 898                         $table = StringUtils::explode( ';', $stripped );
 899                         foreach ( $table as $t ) {
 900                                 $m = explode( '=>', $t, 3 );
 901                                 if ( count( $m ) != 2 )
 902                                         continue;
 903                                 // trim any trailling comments starting with '//'
 904                                 $tt = explode( '//', $m[1], 2 );
 905                                 $ret[trim( $m[0] )] = trim( $tt[0] );
 906                         }
 907                 }
 908                 $parsed[$key] = true;
 909
 910
 911                 // recursively parse the subpages
 912                 if ( $recursive ) {
 913                         foreach ( $sublinks as $link ) {
 914                                 $s = $this->parseCachedTable( $code, $link, $recursive );
 915                                 $ret = array_merge( $ret, $s );
 916                         }
 917                 }
 918
 919                 if ( $this->mUcfirst ) {
 920                         foreach ( $ret as $k => $v ) {
 921                                 $ret[Language::ucfirst( $k )] = Language::ucfirst( $v );
 922                         }
 923                 }
 924                 return $ret;
 925         }
 926
 927         /**
 928          * Enclose a string with the "no conversion" tag. This is used by
 929          * various functions in the Parser.
 930          *
 931          * @param $text String: text to be tagged for no conversion
 932          * @param $noParse Unused (?)
 933          * @return String: the tagged text
 934          */
 935         public function markNoConversion( $text, $noParse = false ) {
 936                 # don't mark if already marked
 937                 if ( strpos( $text, '-{' ) || strpos( $text, '}-' ) ) {
 938                         return $text;
 939                 }
 940
 941                 $ret = "-{R|$text}-";
 942                 return $ret;
 943         }
 944
 945         /**
 946          * Convert the sorting key for category links. This should make different
 947          * keys that are variants of each other map to the same key.
 948          */
 949         function convertCategoryKey( $key ) {
 950                 return $key;
 951         }
 952
 953         /**
 954          * Hook to refresh the cache of conversion tables when
 955          * MediaWiki:conversiontable* is updated.
 956          * @private
 957          */
 958         function OnArticleSaveComplete( $article, $user, $text, $summary, $isminor,
 959                         $iswatch, $section, $flags, $revision ) {
 960                 $titleobj = $article->getTitle();
 961                 if ( $titleobj->getNamespace() == NS_MEDIAWIKI ) {
 962                         $title = $titleobj->getDBkey();
 963                         $t = explode( '/', $title, 3 );
 964                         $c = count( $t );
 965                         if ( $c > 1 && $t[0] == 'Conversiontable' ) {
 966                                 if ( $this->validateVariant( $t[1] ) ) {
 967                                         $this->reloadTables();
 968                                 }
 969                         }
 970                 }
 971                 return true;
 972         }
 973
 974         /**
 975          * Armour rendered math against conversion.
 976          * Wrap math into rawoutput -{R| math }- syntax.
 977          */
 978         public function armourMath( $text ) {
 979                 // we need to convert '-{' and '}-' to '-&#123;' and '&#125;-'
 980                 // to avoid a unwanted '}-' appeared after the math-image.
 981                 $text = strtr( $text, array( '-{' => '-&#123;', '}-' => '&#125;-' ) );
 982                 $ret = "-{R|$text}-";
 983                 return $ret;
 984         }
 985
 986         /**
 987          * Get the cached separator pattern for ConverterRule::parseRules()
 988          */
 989         function getVarSeparatorPattern() {
 990                 if ( is_null( $this->mVarSeparatorPattern ) ) {
 991                         // varsep_pattern for preg_split:
 992                         // text should be splited by ";" only if a valid variant
 993                         // name exist after the markup, for example:
 994                         //  -{zh-hans:<span style="font-size:120%;">xxx</span>;zh-hant:\
 995                         //      <span style="font-size:120%;">yyy</span>;}-
 996                         // we should split it as:
 997                         //  array(
 998                         //        [0] => 'zh-hans:<span style="font-size:120%;">xxx</span>'
 999                         //        [1] => 'zh-hant:<span style="font-size:120%;">yyy</span>'
1000                         //        [2] => ''
1001                         //       )
1002                         $pat = '/;\s*(?=';
1003                         foreach ( $this->mVariants as $variant ) {
1004                                 // zh-hans:xxx;zh-hant:yyy
1005                                 $pat .= $variant . '\s*:|';
1006                                 // xxx=>zh-hans:yyy; xxx=>zh-hant:zzz
1007                                 $pat .= '[^;]*?=>\s*' . $variant . '\s*:|';
1008                         }
1009                         $pat .= '\s*$)/';
1010                         $this->mVarSeparatorPattern = $pat;
1011                 }
1012                 return $this->mVarSeparatorPattern;
1013         }
1014 }
1015
1016 /**
1017  * Parser for rules of language conversion , parse rules in -{ }- tag.
1018  * @ingroup Language
1019  * @author fdcn <fdcn64@gmail.com>, PhiLiP <philip.npc@gmail.com>
1020  */
1021 class ConverterRule {
1022         var $mText; // original text in -{text}-
1023         var $mConverter; // LanguageConverter object
1024         var $mManualCodeError = '<strong class="error">code error!</strong>';
1025         var $mRuleDisplay = '';
1026         var $mRuleTitle = false;
1027         var $mRules = '';// string : the text of the rules
1028         var $mRulesAction = 'none';
1029         var $mFlags = array();
1030         var $mVariantFlags = array();
1031         var $mConvTable = array();
1032         var $mBidtable = array();// array of the translation in each variant
1033         var $mUnidtable = array();// array of the translation in each variant
1034
1035         /**
1036          * Constructor
1037          *
1038          * @param $text String: the text between -{ and }-
1039          * @param $converter LanguageConverter object
1040          */
1041         public function __construct( $text, $converter ) {
1042                 $this->mText = $text;
1043                 $this->mConverter = $converter;
1044         }
1045
1046         /**
1047          * Check if variants array in convert array.
1048          *
1049          * @param $variants Array or string: variant language code
1050          * @return String: translated text
1051          */
1052         public function getTextInBidtable( $variants ) {
1053                 $variants = (array)$variants;
1054                 if ( !$variants ) {
1055                         return false;
1056                 }
1057                 foreach ( $variants as $variant ) {
1058                         if ( isset( $this->mBidtable[$variant] ) ) {
1059                                 return $this->mBidtable[$variant];
1060                         }
1061                 }
1062                 return false;
1063         }
1064
1065         /**
1066          * Parse flags with syntax -{FLAG| ... }-
1067          * @private
1068          */
1069         function parseFlags() {
1070                 $text = $this->mText;
1071                 $flags = array();
1072                 $variantFlags = array();
1073
1074                 $sepPos = strpos( $text, '|' );
1075                 if ( $sepPos !== false ) {
1076                         $validFlags = $this->mConverter->mFlags;
1077                         $f = StringUtils::explode( ';', substr( $text, 0, $sepPos ) );
1078                         foreach ( $f as $ff ) {
1079                                 $ff = trim( $ff );
1080                                 if ( isset( $validFlags[$ff] ) ) {
1081                                         $flags[$validFlags[$ff]] = true;
1082                                 }
1083                         }
1084                         $text = strval( substr( $text, $sepPos + 1 ) );
1085                 }
1086
1087                 if ( !$flags ) {
1088                         $flags['S'] = true;
1089                 } elseif ( isset( $flags['R'] ) ) {
1090                         $flags = array( 'R' => true );// remove other flags
1091                 } elseif ( isset( $flags['N'] ) ) {
1092                         $flags = array( 'N' => true );// remove other flags
1093                 } elseif ( isset( $flags['-'] ) ) {
1094                         $flags = array( '-' => true );// remove other flags
1095                 } elseif ( count( $flags ) == 1 && isset( $flags['T'] ) ) {
1096                         $flags['H'] = true;
1097                 } elseif ( isset( $flags['H'] ) ) {
1098                         // replace A flag, and remove other flags except T
1099                         $temp = array( '+' => true, 'H' => true );
1100                         if ( isset( $flags['T'] ) ) {
1101                                 $temp['T'] = true;
1102                         }
1103                         if ( isset( $flags['D'] ) ) {
1104                                 $temp['D'] = true;
1105                         }
1106                         $flags = $temp;
1107                 } else {
1108                         if ( isset( $flags['A'] ) ) {
1109                                 $flags['+'] = true;
1110                                 $flags['S'] = true;
1111                         }
1112                         if ( isset( $flags['D'] ) ) {
1113                                 unset( $flags['S'] );
1114                         }
1115                         // try to find flags like "zh-hans", "zh-hant"
1116                         // allow syntaxes like "-{zh-hans;zh-hant|XXXX}-"
1117                         $variantFlags = array_intersect( array_keys( $flags ), $this->mConverter->mVariants );
1118                         if ( $variantFlags ) {
1119                                 $variantFlags = array_flip( $variantFlags );
1120                                 $flags = array();
1121                         }
1122                 }
1123                 $this->mVariantFlags = $variantFlags;
1124                 $this->mRules = $text;
1125                 $this->mFlags = $flags;
1126         }
1127
1128         /**
1129          * Generate conversion table.
1130          * @private
1131          */
1132         function parseRules() {
1133                 $rules = $this->mRules;
1134                 $flags = $this->mFlags;
1135                 $bidtable = array();
1136                 $unidtable = array();
1137                 $variants = $this->mConverter->mVariants;
1138                 $varsep_pattern = $this->mConverter->getVarSeparatorPattern();
1139
1140                 $choice = preg_split( $varsep_pattern, $rules );
1141
1142                 foreach ( $choice as $c ) {
1143                         $v  = explode( ':', $c, 2 );
1144                         if ( count( $v ) != 2 ) {
1145                                 // syntax error, skip
1146                                 continue;
1147                         }
1148                         $to = trim( $v[1] );
1149                         $v  = trim( $v[0] );
1150                         $u  = explode( '=>', $v, 2 );
1151                         // if $to is empty, strtr() could return a wrong result
1152                         if ( count( $u ) == 1 && $to && in_array( $v, $variants ) ) {
1153                                 $bidtable[$v] = $to;
1154                         } elseif ( count( $u ) == 2 ) {
1155                                 $from = trim( $u[0] );
1156                                 $v      = trim( $u[1] );
1157                                 if ( array_key_exists( $v, $unidtable )
1158                                          && !is_array( $unidtable[$v] )
1159                                          && $to
1160                                          && in_array( $v, $variants ) ) {
1161                                         $unidtable[$v] = array( $from => $to );
1162                                 } elseif ( $to && in_array( $v, $variants ) ) {
1163                                         $unidtable[$v][$from] = $to;
1164                                 }
1165                         }
1166                         // syntax error, pass
1167                         if ( !isset( $this->mConverter->mVariantNames[$v] ) ) {
1168                                 $bidtable = array();
1169                                 $unidtable = array();
1170                                 break;
1171                         }
1172                 }
1173                 $this->mBidtable = $bidtable;
1174                 $this->mUnidtable = $unidtable;
1175         }
1176
1177         /**
1178          * @private
1179          */
1180         function getRulesDesc() {
1181                 $codesep = $this->mConverter->mDescCodeSep;
1182                 $varsep = $this->mConverter->mDescVarSep;
1183                 $text = '';
1184                 foreach ( $this->mBidtable as $k => $v ) {
1185                         $text .= $this->mConverter->mVariantNames[$k] . "$codesep$v$varsep";
1186                 }
1187                 foreach ( $this->mUnidtable as $k => $a ) {
1188                         foreach ( $a as $from => $to ) {
1189                                 $text .= $from . '⇒' . $this->mConverter->mVariantNames[$k] .
1190                                         "$codesep$to$varsep";
1191                         }
1192                 }
1193                 return $text;
1194         }
1195
1196         /**
1197          * Parse rules conversion.
1198          * @private
1199          */
1200         function getRuleConvertedStr( $variant ) {
1201                 $bidtable = $this->mBidtable;
1202                 $unidtable = $this->mUnidtable;
1203
1204                 if ( count( $bidtable ) + count( $unidtable ) == 0 ) {
1205                         return $this->mRules;
1206                 } else {
1207                         // display current variant in bidirectional array
1208                         $disp = $this->getTextInBidtable( $variant );
1209                         // or display current variant in fallbacks
1210                         if ( !$disp ) {
1211                                 $disp = $this->getTextInBidtable(
1212                                                 $this->mConverter->getVariantFallbacks( $variant ) );
1213                         }
1214                         // or display current variant in unidirectional array
1215                         if ( !$disp && array_key_exists( $variant, $unidtable ) ) {
1216                                 $disp = array_values( $unidtable[$variant] );
1217                                 $disp = $disp[0];
1218                         }
1219                         // or display frist text under disable manual convert
1220                         if ( !$disp
1221                                  && $this->mConverter->mManualLevel[$variant] == 'disable' ) {
1222                                 if ( count( $bidtable ) > 0 ) {
1223                                         $disp = array_values( $bidtable );
1224                                         $disp = $disp[0];
1225                                 } else {
1226                                         $disp = array_values( $unidtable );
1227                                         $disp = array_values( $disp[0] );
1228                                         $disp = $disp[0];
1229                                 }
1230                         }
1231                         return $disp;
1232                 }
1233         }
1234
1235         /**
1236          * Generate conversion table for all text.
1237          * @private
1238          */
1239         function generateConvTable() {
1240                 // Special case optimisation
1241                 if ( !$this->mBidtable && !$this->mUnidtable ) {
1242                         $this->mConvTable = array();
1243                         return;
1244                 }
1245
1246                 $bidtable = $this->mBidtable;
1247                 $unidtable = $this->mUnidtable;
1248                 $manLevel = $this->mConverter->mManualLevel;
1249
1250                 $vmarked = array();
1251                 foreach ( $this->mConverter->mVariants as $v ) {
1252                         /* for bidirectional array
1253                                 fill in the missing variants, if any,
1254                                 with fallbacks */
1255                         if ( !isset( $bidtable[$v] ) ) {
1256                                 $variantFallbacks =
1257                                         $this->mConverter->getVariantFallbacks( $v );
1258                                 $vf = $this->getTextInBidtable( $variantFallbacks );
1259                                 if ( $vf ) {
1260                                         $bidtable[$v] = $vf;
1261                                 }
1262                         }
1263
1264                         if ( isset( $bidtable[$v] ) ) {
1265                                 foreach ( $vmarked as $vo ) {
1266                                         // use syntax: -{A|zh:WordZh;zh-tw:WordTw}-
1267                                         // or -{H|zh:WordZh;zh-tw:WordTw}-
1268                                         // or -{-|zh:WordZh;zh-tw:WordTw}-
1269                                         // to introduce a custom mapping between
1270                                         // words WordZh and WordTw in the whole text
1271                                         if ( $manLevel[$v] == 'bidirectional' ) {
1272                                                 $this->mConvTable[$v][$bidtable[$vo]] = $bidtable[$v];
1273                                         }
1274                                         if ( $manLevel[$vo] == 'bidirectional' ) {
1275                                                 $this->mConvTable[$vo][$bidtable[$v]] = $bidtable[$vo];
1276                                         }
1277                                 }
1278                                 $vmarked[] = $v;
1279                         }
1280                         /*for unidirectional array fill to convert tables */
1281                         if ( ( $manLevel[$v] == 'bidirectional' || $manLevel[$v] == 'unidirectional' )
1282                                 && isset( $unidtable[$v] ) )
1283                         {
1284                                 if ( isset( $this->mConvTable[$v] ) ) {
1285                                         $this->mConvTable[$v] = array_merge( $this->mConvTable[$v], $unidtable[$v] );
1286                                 } else {
1287                                         $this->mConvTable[$v] = $unidtable[$v];
1288                                 }
1289                         }
1290                 }
1291         }
1292
1293         /**
1294          * Parse rules and flags.
1295          * @public
1296          */
1297         function parse( $variant = NULL ) {
1298                 if ( !$variant ) {
1299                         $variant = $this->mConverter->getPreferredVariant();
1300                 }
1301
1302                 $variants = $this->mConverter->mVariants;
1303                 $this->parseFlags();
1304                 $flags = $this->mFlags;
1305
1306                 // convert to specified variant
1307                 // syntax: -{zh-hans;zh-hant[;...]|<text to convert>}-
1308                 if ( $this->mVariantFlags ) {
1309                         // check if current variant in flags
1310                         if ( isset( $this->mVariantFlags[$variant] ) ) {
1311                                 // then convert <text to convert> to current language
1312                                 $this->mRules = $this->mConverter->autoConvert( $this->mRules,
1313                                                                                                                                 $variant );
1314                         } else { // if current variant no in flags,
1315                                    // then we check its fallback variants.
1316                                 $variantFallbacks =
1317                                         $this->mConverter->getVariantFallbacks( $variant );
1318                                 foreach ( $variantFallbacks as $variantFallback ) {
1319                                         // if current variant's fallback exist in flags
1320                                         if ( isset( $this->mVariantFlags[$variantFallback] ) ) {
1321                                                 // then convert <text to convert> to fallback language
1322                                                 $this->mRules =
1323                                                         $this->mConverter->autoConvert( $this->mRules,
1324                                                                                                                         $variantFallback );
1325                                                 break;
1326                                         }
1327                                 }
1328                         }
1329                         $this->mFlags = $flags = array( 'R' => true );
1330                 }
1331
1332                 if ( !isset( $flags['R'] ) && !isset( $flags['N'] ) ) {
1333                         // decode => HTML entities modified by Sanitizer::removeHTMLtags
1334                         $this->mRules = str_replace( '=&gt;', '=>', $this->mRules );
1335                         $this->parseRules();
1336                 }
1337                 $rules = $this->mRules;
1338
1339                 if ( !$this->mBidtable && !$this->mUnidtable ) {
1340                         if ( isset( $flags['+'] ) || isset( $flags['-'] ) ) {
1341                                 // fill all variants if text in -{A/H/-|text} without rules
1342                                 foreach ( $this->mConverter->mVariants as $v ) {
1343                                         $this->mBidtable[$v] = $rules;
1344                                 }
1345                         } elseif ( !isset( $flags['N'] ) && !isset( $flags['T'] ) ) {
1346                                 $this->mFlags = $flags = array( 'R' => true );
1347                         }
1348                 }
1349
1350                 $this->mRuleDisplay = false;
1351                 foreach ( $flags as $flag => $unused ) {
1352                         switch ( $flag ) {
1353                                 case 'R':
1354                                         // if we don't do content convert, still strip the -{}- tags
1355                                         $this->mRuleDisplay = $rules;
1356                                         break;
1357                                 case 'N':
1358                                         // process N flag: output current variant name
1359                                         $ruleVar = trim( $rules );
1360                                         if ( isset( $this->mConverter->mVariantNames[$ruleVar] ) ) {
1361                                                 $this->mRuleDisplay = $this->mConverter->mVariantNames[$ruleVar];
1362                                         } else {
1363                                                 $this->mRuleDisplay = '';
1364                                         }
1365                                         break;
1366                                 case 'D':
1367                                         // process D flag: output rules description
1368                                         $this->mRuleDisplay = $this->getRulesDesc();
1369                                         break;
1370                                 case 'H':
1371                                         // process H,- flag or T only: output nothing
1372                                         $this->mRuleDisplay = '';
1373                                         break;
1374                                 case '-':
1375                                         $this->mRulesAction = 'remove';
1376                                         $this->mRuleDisplay = '';
1377                                         break;
1378                                 case '+':
1379                                         $this->mRulesAction = 'add';
1380                                         $this->mRuleDisplay = '';
1381                                         break;
1382                                 case 'S':
1383                                         $this->mRuleDisplay = $this->getRuleConvertedStr( $variant );
1384                                         break;
1385                                 case 'T':
1386                                         $this->mRuleTitle = $this->getRuleConvertedStr( $variant );
1387                                         $this->mRuleDisplay = '';
1388                                         break;
1389                                 default:
1390                                         // ignore unknown flags (but see error case below)
1391                         }
1392                 }
1393                 if ( $this->mRuleDisplay === false ) {
1394                         $this->mRuleDisplay = $this->mManualCodeError;
1395                 }
1396
1397                 $this->generateConvTable();
1398         }
1399
1400         /**
1401          * @public
1402          */
1403         function hasRules() {
1404                 // TODO:
1405         }
1406
1407         /**
1408          * Get display text on markup -{...}-
1409          * @public
1410          */
1411         function getDisplay() {
1412                 return $this->mRuleDisplay;
1413         }
1414
1415         /**
1416          * Get converted title.
1417          * @public
1418          */
1419         function getTitle() {
1420                 return $this->mRuleTitle;
1421         }
1422
1423         /**
1424          * Return how deal with conversion rules.
1425          * @public
1426          */
1427         function getRulesAction() {
1428                 return $this->mRulesAction;
1429         }
1430
1431         /**
1432          * Get conversion table. ( bidirectional and unidirectional
1433          * conversion table )
1434          * @public
1435          */
1436         function getConvTable() {
1437                 return $this->mConvTable;
1438         }
1439
1440         /**
1441          * Get conversion rules string.
1442          * @public
1443          */
1444         function getRules() {
1445                 return $this->mRules;
1446         }
1447
1448         /**
1449          * Get conversion flags.
1450          * @public
1451          */
1452         function getFlags() {
1453                 return $this->mFlags;
1454         }
1455 }