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
  39         /**
  40          * @var String: memcached key name
  41          */
  42         var $mCacheKey;
  43
  44         var $mLangObj;
  45         var $mFlags;
  46         var $mDescCodeSep = ':', $mDescVarSep = ';';
  47         var $mUcfirst = false;
  48         var $mConvRuleTitle = false;
  49         var $mURLVariant;
  50         var $mUserVariant;
  51         var $mHeaderVariant;
  52         var $mMaxDepth = 10;
  53         var $mVarSeparatorPattern;
  54
  55         const CACHE_VERSION_KEY = 'VERSION 6';
  56
  57         /**
  58          * Constructor
  59          *
  60          * @param $langobj Language: the Language Object
  61          * @param $maincode String: the main language code of this language
  62          * @param $variants Array: the supported variants of this language
  63          * @param $variantfallbacks Array: the fallback language of each variant
  64          * @param $flags Array: defining the custom strings that maps to the flags
  65          * @param $manualLevel Array: limit for supported variants
  66          */
  67         public function __construct( $langobj, $maincode, $variants = array(),
  68                                                                 $variantfallbacks = array(), $flags = array(),
  69                                                                 $manualLevel = array() ) {
  70                 global $wgDisabledVariants;
  71                 $this->mLangObj = $langobj;
  72                 $this->mMainLanguageCode = $maincode;
  73                 $this->mVariants = array_diff( $variants, $wgDisabledVariants );
  74                 $this->mVariantFallbacks = $variantfallbacks;
  75                 $this->mVariantNames = Language::getLanguageNames();
  76                 $this->mCacheKey = wfMemcKey( 'conversiontables', $maincode );
  77                 $defaultflags = array(
  78                         // 'S' show converted text
  79                         // '+' add rules for alltext
  80                         // 'E' the gave flags is error
  81                         // these flags above are reserved for program
  82                         'A' => 'A',       // add rule for convert code (all text convert)
  83                         'T' => 'T',       // title convert
  84                         'R' => 'R',       // raw content
  85                         'D' => 'D',       // convert description (subclass implement)
  86                         '-' => '-',       // remove convert (not implement)
  87                         'H' => 'H',       // add rule for convert code
  88                                                   // (but no display in placed code)
  89                         'N' => 'N'        // current variant name
  90                 );
  91                 $this->mFlags = array_merge( $defaultflags, $flags );
  92                 foreach ( $this->mVariants as $v ) {
  93                         if ( array_key_exists( $v, $manualLevel ) ) {
  94                                 $this->mManualLevel[$v] = $manualLevel[$v];
  95                         } else {
  96                                 $this->mManualLevel[$v] = 'bidirectional';
  97                         }
  98                         $this->mFlags[$v] = $v;
  99                 }
 100         }
 101
 102         /**
 103          * Get all valid variants.
 104          * Call this instead of using $this->mVariants directly.
 105          *
 106          * @return Array: contains all valid variants
 107          */
 108         public function getVariants() {
 109                 return $this->mVariants;
 110         }
 111
 112         /**
 113          * In case some variant is not defined in the markup, we need
 114          * to have some fallback. For example, in zh, normally people
 115          * will define zh-hans and zh-hant, but less so for zh-sg or zh-hk.
 116          * when zh-sg is preferred but not defined, we will pick zh-hans
 117          * in this case. Right now this is only used by zh.
 118          *
 119          * @param $variant String: the language code of the variant
 120          * @return String: The code of the fallback language or the
 121          *                               main code if there is no fallback
 122          */
 123         public function getVariantFallbacks( $variant ) {
 124                 if ( isset( $this->mVariantFallbacks[$variant] ) ) {
 125                         return $this->mVariantFallbacks[$variant];
 126                 }
 127                 return $this->mMainLanguageCode;
 128         }
 129
 130         /**
 131          * Get the title produced by the conversion rule.
 132          * @return String: The converted title text
 133          */
 134         public function getConvRuleTitle() {
 135                 return $this->mConvRuleTitle;
 136         }
 137
 138         /**
 139          * Get preferred language variant.
 140          * @return String: the preferred language code
 141          */
 142         public function getPreferredVariant() {
 143                 global $wgDefaultLanguageVariant, $wgUser;
 144
 145                 $req = $this->getURLVariant();
 146
 147                 if ( $wgUser->isLoggedIn() && !$req ) {
 148                         $req = $this->getUserVariant();
 149                 } elseif ( !$req ) {
 150                         $req = $this->getHeaderVariant();
 151                 }
 152
 153                 if ( $wgDefaultLanguageVariant && !$req ) {
 154                         $req = $this->validateVariant( $wgDefaultLanguageVariant );
 155                 }
 156
 157                 // This function, unlike the other get*Variant functions, is
 158                 // not memoized (i.e. there return value is not cached) since
 159                 // new information might appear during processing after this
 160                 // is first called.
 161                 if ( $req ) {
 162                         return $req;
 163                 }
 164                 return $this->mMainLanguageCode;
 165         }
 166
 167         /**
 168          * Get default variant.
 169          * This function would not be affected by user's settings or headers
 170          * @return String: the default variant code
 171          */
 172         public function getDefaultVariant() {
 173                 global $wgDefaultLanguageVariant;
 174
 175                 $req = $this->getURLVariant();
 176
 177                 if ( $wgDefaultLanguageVariant && !$req ) {
 178                         $req = $this->validateVariant( $wgDefaultLanguageVariant );
 179                 }
 180
 181                 if ( $req ) {
 182                         return $req;
 183                 }
 184                 return $this->mMainLanguageCode;
 185         }
 186
 187         /**
 188          * Validate the variant
 189          * @param $variant String: the variant to validate
 190          * @return Mixed: returns the variant if it is valid, null otherwise
 191          */
 192         protected function validateVariant( $variant = null ) {
 193                 if ( $variant !== null && in_array( $variant, $this->mVariants ) ) {
 194                         return $variant;
 195                 }
 196                 return null;
 197         }
 198
 199         /**
 200          * Get the variant specified in the URL
 201          *
 202          * @return Mixed: variant if one found, false otherwise.
 203          */
 204         public function getURLVariant() {
 205                 global $wgRequest;
 206
 207                 if ( $this->mURLVariant ) {
 208                         return $this->mURLVariant;
 209                 }
 210
 211                 // see if the preference is set in the request
 212                 $ret = $wgRequest->getText( 'variant' );
 213
 214                 if ( !$ret ) {
 215                         $ret = $wgRequest->getVal( 'uselang' );
 216                 }
 217
 218                 return $this->mURLVariant = $this->validateVariant( $ret );
 219         }
 220
 221         /**
 222          * Determine if the user has a variant set.
 223          *
 224          * @return Mixed: variant if one found, false otherwise.
 225          */
 226         protected function getUserVariant() {
 227                 global $wgUser;
 228
 229                 // memoizing this function wreaks havoc on parserTest.php
 230                 /*
 231                 if ( $this->mUserVariant ) {
 232                         return $this->mUserVariant;
 233                 }
 234                 */
 235
 236                 // Get language variant preference from logged in users
 237                 // Don't call this on stub objects because that causes infinite
 238                 // recursion during initialisation
 239                 if ( $wgUser->isLoggedIn() )  {
 240                         $ret = $wgUser->getOption( 'variant' );
 241                 } else {
 242                         // figure out user lang without constructing wgLang to avoid
 243                         // infinite recursion
 244                         $ret = $wgUser->getOption( 'language' );
 245                 }
 246
 247                 return $this->mUserVariant = $this->validateVariant( $ret );
 248         }
 249
 250         /**
 251          * Determine the language variant from the Accept-Language header.
 252          *
 253          * @return Mixed: variant if one found, false otherwise.
 254          */
 255         protected function getHeaderVariant() {
 256                 global $wgRequest;
 257
 258                 if ( $this->mHeaderVariant ) {
 259                         return $this->mHeaderVariant;
 260                 }
 261
 262                 // see if some supported language variant is set in the
 263                 // HTTP header.
 264                 $languages = array_keys( $wgRequest->getAcceptLang() );
 265                 if ( empty( $languages ) ) {
 266                         return null;
 267                 }
 268
 269                 $fallbackLanguages = array();
 270                 foreach ( $languages as $language ) {
 271                         $this->mHeaderVariant = $this->validateVariant( $language );
 272                         if ( $this->mHeaderVariant ) {
 273                                 break;
 274                         }
 275
 276                         // To see if there are fallbacks of current language.
 277                         // We record these fallback variants, and process
 278                         // them later.
 279                         $fallbacks = $this->getVariantFallbacks( $language );
 280                         if ( is_string( $fallbacks ) ) {
 281                                 $fallbackLanguages[] = $fallbacks;
 282                         } elseif ( is_array( $fallbacks ) ) {
 283                                 $fallbackLanguages =
 284                                         array_merge( $fallbackLanguages, $fallbacks );
 285                         }
 286                 }
 287
 288                 if ( !$this->mHeaderVariant ) {
 289                         // process fallback languages now
 290                         $fallback_languages = array_unique( $fallbackLanguages );
 291                         foreach ( $fallback_languages as $language ) {
 292                                 $this->mHeaderVariant = $this->validateVariant( $language );
 293                                 if ( $this->mHeaderVariant ) {
 294                                         break;
 295                                 }
 296                         }
 297                 }
 298
 299                 return $this->mHeaderVariant;
 300         }
 301
 302         /**
 303          * Dictionary-based conversion.
 304          * This function would not parse the conversion rules.
 305          * If you want to parse rules, try to use convert() or
 306          * convertTo().
 307          *
 308          * @param $text String: the text to be converted
 309          * @param $toVariant String: the target language code
 310          * @return String: the converted text
 311          */
 312         public function autoConvert( $text, $toVariant = false ) {
 313                 wfProfileIn( __METHOD__ );
 314
 315                 $this->loadTables();
 316
 317                 if ( !$toVariant ) {
 318                         $toVariant = $this->getPreferredVariant();
 319                         if ( !$toVariant ) {
 320                                 wfProfileOut( __METHOD__ );
 321                                 return $text;
 322                         }
 323                 }
 324
 325                 /* we convert everything except:
 326                    1. HTML markups (anything between < and >)
 327                    2. HTML entities
 328                    3. placeholders created by the parser
 329                 */
 330                 global $wgParser;
 331                 if ( isset( $wgParser ) && $wgParser->UniqPrefix() != '' ) {
 332                         $marker = '|' . $wgParser->UniqPrefix() . '[\-a-zA-Z0-9]+';
 333                 } else {
 334                         $marker = '';
 335                 }
 336
 337                 // this one is needed when the text is inside an HTML markup
 338                 $htmlfix = '|<[^>]+$|^[^<>]*>';
 339
 340                 // disable convert to variants between <code></code> tags
 341                 $codefix = '<code>.+?<\/code>|';
 342                 // disable convertsion of <script type="text/javascript"> ... </script>
 343                 $scriptfix = '<script.*?>.*?<\/script>|';
 344                 // disable conversion of <pre xxxx> ... </pre>
 345                 $prefix = '<pre.*?>.*?<\/pre>|';
 346
 347                 $reg = '/' . $codefix . $scriptfix . $prefix .
 348                         '<[^>]+>|&[a-zA-Z#][a-z0-9]+;' . $marker . $htmlfix . '/s';
 349                 $startPos = 0;
 350                 $sourceBlob = '';
 351                 $literalBlob = '';
 352
 353                 // Guard against delimiter nulls in the input
 354                 $text = str_replace( "\000", '', $text );
 355
 356                 $markupMatches = null;
 357                 $elementMatches = null;
 358                 while ( $startPos < strlen( $text ) ) {
 359                         if ( preg_match( $reg, $text, $markupMatches, PREG_OFFSET_CAPTURE, $startPos ) ) {
 360                                 $elementPos = $markupMatches[0][1];
 361                                 $element = $markupMatches[0][0];
 362                         } else {
 363                                 $elementPos = strlen( $text );
 364                                 $element = '';
 365                         }
 366
 367                         // Queue the part before the markup for translation in a batch
 368                         $sourceBlob .= substr( $text, $startPos, $elementPos - $startPos ) . "\000";
 369
 370                         // Advance to the next position
 371                         $startPos = $elementPos + strlen( $element );
 372
 373                         // Translate any alt or title attributes inside the matched element
 374                         if ( $element !== '' && preg_match( '/^(<[^>\s]*)\s([^>]*)(.*)$/', $element,
 375                                 $elementMatches ) )
 376                         {
 377                                 $attrs = Sanitizer::decodeTagAttributes( $elementMatches[2] );
 378                                 $changed = false;
 379                                 foreach ( array( 'title', 'alt' ) as $attrName ) {
 380                                         if ( !isset( $attrs[$attrName] ) ) {
 381                                                 continue;
 382                                         }
 383                                         $attr = $attrs[$attrName];
 384                                         // Don't convert URLs
 385                                         if ( !strpos( $attr, '://' ) ) {
 386                                                 $attr = $this->translate( $attr, $toVariant );
 387                                         }
 388
 389                                         // Remove HTML tags to avoid disrupting the layout
 390                                         $attr = preg_replace( '/<[^>]+>/', '', $attr );
 391                                         if ( $attr !== $attrs[$attrName] ) {
 392                                                 $attrs[$attrName] = $attr;
 393                                                 $changed = true;
 394                                         }
 395                                 }
 396                                 if ( $changed ) {
 397                                         $element = $elementMatches[1] . Html::expandAttributes( $attrs ) .
 398                                                 $elementMatches[3];
 399                                 }
 400                         }
 401                         $literalBlob .= $element . "\000";
 402                 }
 403
 404                 // Do the main translation batch
 405                 $translatedBlob = $this->translate( $sourceBlob, $toVariant );
 406
 407                 // Put the output back together
 408                 $translatedIter = StringUtils::explode( "\000", $translatedBlob );
 409                 $literalIter = StringUtils::explode( "\000", $literalBlob );
 410                 $output = '';
 411                 while ( $translatedIter->valid() && $literalIter->valid() ) {
 412                         $output .= $translatedIter->current();
 413                         $output .= $literalIter->current();
 414                         $translatedIter->next();
 415                         $literalIter->next();
 416                 }
 417
 418                 wfProfileOut( __METHOD__ );
 419                 return $output;
 420         }
 421
 422         /**
 423          * Translate a string to a variant.
 424          * Doesn't parse rules or do any of that other stuff, for that use
 425          * convert() or convertTo().
 426          *
 427          * @param $text String: text to convert
 428          * @param $variant String: variant language code
 429          * @return String: translated text
 430          */
 431         public function translate( $text, $variant ) {
 432                 wfProfileIn( __METHOD__ );
 433                 // If $text is empty or only includes spaces, do nothing
 434                 // Otherwise translate it
 435                 if ( trim( $text ) ) {
 436                         $this->loadTables();
 437                         $text = $this->mTables[$variant]->replace( $text );
 438                 }
 439                 wfProfileOut( __METHOD__ );
 440                 return $text;
 441         }
 442
 443         /**
 444          * Call translate() to convert text to all valid variants.
 445          *
 446          * @param $text String: the text to be converted
 447          * @return Array: variant => converted text
 448          */
 449         public function autoConvertToAllVariants( $text ) {
 450                 wfProfileIn( __METHOD__ );
 451                 $this->loadTables();
 452
 453                 $ret = array();
 454                 foreach ( $this->mVariants as $variant ) {
 455                         $ret[$variant] = $this->translate( $text, $variant );
 456                 }
 457
 458                 wfProfileOut( __METHOD__ );
 459                 return $ret;
 460         }
 461
 462         /**
 463          * Convert link text to all valid variants.
 464          * In the first, this function only convert text outside the
 465          * "-{" "}-" markups. Since the "{" and "}" are not allowed in
 466          * titles, the text will get all converted always.
 467          * So I removed this feature and deprecated the function.
 468          *
 469          * @param $text String: the text to be converted
 470          * @return Array: variant => converted text
 471          * @deprecated Use autoConvertToAllVariants() instead
 472          */
 473         public function convertLinkToAllVariants( $text ) {
 474                 return $this->autoConvertToAllVariants( $text );
 475         }
 476
 477         /**
 478          * Apply manual conversion rules.
 479          *
 480          * @param $convRule Object: Object of ConverterRule
 481          */
 482         protected function applyManualConv( $convRule ) {
 483                 // Use syntax -{T|zh-cn:TitleCN; zh-tw:TitleTw}- to custom
 484                 // title conversion.
 485                 // Bug 24072: $mConvRuleTitle was overwritten by other manual
 486                 // rule(s) not for title, this breaks the title conversion.
 487                 $newConvRuleTitle = $convRule->getTitle();
 488                 if ( $newConvRuleTitle ) {
 489                         // So I add an empty check for getTitle()
 490                         $this->mConvRuleTitle = $newConvRuleTitle;
 491                 }
 492
 493                 // merge/remove manual conversion rules to/from global table
 494                 $convTable = $convRule->getConvTable();
 495                 $action = $convRule->getRulesAction();
 496                 foreach ( $convTable as $variant => $pair ) {
 497                         if ( !$this->validateVariant( $variant ) ) {
 498                                 continue;
 499                         }
 500
 501                         if ( $action == 'add' ) {
 502                                 foreach ( $pair as $from => $to ) {
 503                                         // to ensure that $from and $to not be left blank
 504                                         // so $this->translate() could always return a string
 505                                         if ( $from || $to ) {
 506                                                 // more efficient than array_merge(), about 2.5 times.
 507                                                 $this->mTables[$variant]->setPair( $from, $to );
 508                                         }
 509                                 }
 510                         } elseif ( $action == 'remove' ) {
 511                                 $this->mTables[$variant]->removeArray( $pair );
 512                         }
 513                 }
 514         }
 515
 516         /**
 517          * Auto convert a Title object to a readable string in the
 518          * preferred variant.
 519          *
 520          * @param $title Object: a object of Title
 521          * @return String: converted title text
 522          */
 523         public function convertTitle( $title ) {
 524                 $variant = $this->getPreferredVariant();
 525                 $index = $title->getNamespace();
 526                 if ( $index === NS_MAIN ) {
 527                         $text = '';
 528                 } else {
 529                         // first let's check if a message has given us a converted name
 530                         $nsConvKey = 'conversion-ns' . $index;
 531                         if ( !wfEmptyMsg( $nsConvKey ) ) {
 532                                 $text = wfMsgForContentNoTrans( $nsConvKey );
 533                         } else {
 534                                 // the message does not exist, try retrieve it from the current
 535                                 // variant's namespace names.
 536                                 $langObj = $this->mLangObj->factory( $variant );
 537                                 $text = $langObj->getFormattedNsText( $index );
 538                         }
 539                         $text .= ':';
 540                 }
 541                 $text .= $title->getText();
 542                 $text = $this->translate( $text, $variant );
 543                 return $text;
 544         }
 545
 546         /**
 547          * Convert text to different variants of a language. The automatic
 548          * conversion is done in autoConvert(). Here we parse the text
 549          * marked with -{}-, which specifies special conversions of the
 550          * text that can not be accomplished in autoConvert().
 551          *
 552          * Syntax of the markup:
 553          * -{code1:text1;code2:text2;...}-  or
 554          * -{flags|code1:text1;code2:text2;...}-  or
 555          * -{text}- in which case no conversion should take place for text
 556          *
 557          * @param $text String: text to be converted
 558          * @return String: converted text
 559          */
 560         public function convert( $text ) {
 561                 $variant = $this->getPreferredVariant();
 562                 return $this->convertTo( $text, $variant );
 563         }
 564
 565         /**
 566          * Same as convert() except a extra parameter to custom variant.
 567          *
 568          * @param $text String: text to be converted
 569          * @param $variant String: the target variant code
 570          * @return String: converted text
 571          */
 572         public function convertTo( $text, $variant ) {
 573                 global $wgDisableLangConversion;
 574                 if ( $wgDisableLangConversion ) {
 575                         return $text;
 576                 }
 577                 return $this->recursiveConvertTopLevel( $text, $variant );
 578         }
 579
 580         /**
 581          * Recursively convert text on the outside. Allow to use nested
 582          * markups to custom rules.
 583          *
 584          * @param $text String: text to be converted
 585          * @param $variant String: the target variant code
 586          * @param $depth Integer: depth of recursion
 587          * @return String: converted text
 588          */
 589         protected function recursiveConvertTopLevel( $text, $variant, $depth = 0 ) {
 590                 $startPos = 0;
 591                 $out = '';
 592                 $length = strlen( $text );
 593                 while ( $startPos < $length ) {
 594                         $pos = strpos( $text, '-{', $startPos );
 595
 596                         if ( $pos === false ) {
 597                                 // No more markup, append final segment
 598                                 $out .= $this->autoConvert( substr( $text, $startPos ), $variant );
 599                                 return $out;
 600                         }
 601
 602                         // Markup found
 603                         // Append initial segment
 604                         $out .= $this->autoConvert( substr( $text, $startPos, $pos - $startPos ), $variant );
 605
 606                         // Advance position
 607                         $startPos = $pos;
 608
 609                         // Do recursive conversion
 610                         $out .= $this->recursiveConvertRule( $text, $variant, $startPos, $depth + 1 );
 611                 }
 612
 613                 return $out;
 614         }
 615
 616         /**
 617          * Recursively convert text on the inside.
 618          *
 619          * @param $text String: text to be converted
 620          * @param $variant String: the target variant code
 621          * @param $depth Integer: depth of recursion
 622          * @return String: converted text
 623          */
 624         protected function recursiveConvertRule( $text, $variant, &$startPos, $depth = 0 ) {
 625                 // Quick sanity check (no function calls)
 626                 if ( $text[$startPos] !== '-' || $text[$startPos + 1] !== '{' ) {
 627                         throw new MWException( __METHOD__ . ': invalid input string' );
 628                 }
 629
 630                 $startPos += 2;
 631                 $inner = '';
 632                 $warningDone = false;
 633                 $length = strlen( $text );
 634
 635                 while ( $startPos < $length ) {
 636                         $m = false;
 637                         preg_match( '/-\{|\}-/', $text, $m,  PREG_OFFSET_CAPTURE, $startPos );
 638                         if ( !$m ) {
 639                                 // Unclosed rule
 640                                 break;
 641                         }
 642
 643                         $token = $m[0][0];
 644                         $pos = $m[0][1];
 645
 646                         // Markup found
 647                         // Append initial segment
 648                         $inner .= substr( $text, $startPos, $pos - $startPos );
 649
 650                         // Advance position
 651                         $startPos = $pos;
 652
 653                         switch ( $token ) {
 654                                 case '-{':
 655                                         // Check max depth
 656                                         if ( $depth >= $this->mMaxDepth ) {
 657                                                 $inner .= '-{';
 658                                                 if ( !$warningDone ) {
 659                                                         $inner .= '<span class="error">' .
 660                                                                 wfMsgForContent( 'language-converter-depth-warning',
 661                                                                         $this->mMaxDepth ) .
 662                                                                 '</span>';
 663                                                         $warningDone = true;
 664                                                 }
 665                                                 $startPos += 2;
 666                                                 continue;
 667                                         }
 668                                         // Recursively parse another rule
 669                                         $inner .= $this->recursiveConvertRule( $text, $variant, $startPos, $depth + 1 );
 670                                         break;
 671                                 case '}-':
 672                                         // Apply the rule
 673                                         $startPos += 2;
 674                                         $rule = new ConverterRule( $inner, $this );
 675                                         $rule->parse( $variant );
 676                                         $this->applyManualConv( $rule );
 677                                         return $rule->getDisplay();
 678                                 default:
 679                                         throw new MWException( __METHOD__ . ': invalid regex match' );
 680                         }
 681                 }
 682
 683                 // Unclosed rule
 684                 if ( $startPos < $length ) {
 685                         $inner .= substr( $text, $startPos );
 686                 }
 687                 $startPos = $length;
 688                 return '-{' . $this->autoConvert( $inner, $variant );
 689         }
 690
 691         /**
 692          * If a language supports multiple variants, it is possible that
 693          * non-existing link in one variant actually exists in another variant.
 694          * This function tries to find it. See e.g. LanguageZh.php
 695          *
 696          * @param $link String: the name of the link
 697          * @param $nt Mixed: the title object of the link
 698          * @param $ignoreOtherCond Boolean: to disable other conditions when
 699          *              we need to transclude a template or update a category's link
 700          * @return Null, the input parameters may be modified upon return
 701          */
 702         public function findVariantLink( &$link, &$nt, $ignoreOtherCond = false ) {
 703                 # If the article has already existed, there is no need to
 704                 # check it again, otherwise it may cause a fault.
 705                 if ( is_object( $nt ) && $nt->exists() ) {
 706                         return;
 707                 }
 708
 709                 global $wgDisableLangConversion, $wgDisableTitleConversion, $wgRequest,
 710                         $wgUser;
 711                 $isredir = $wgRequest->getText( 'redirect', 'yes' );
 712                 $action = $wgRequest->getText( 'action' );
 713                 $linkconvert = $wgRequest->getText( 'linkconvert', 'yes' );
 714                 $disableLinkConversion = $wgDisableLangConversion
 715                         || $wgDisableTitleConversion;
 716                 $linkBatch = new LinkBatch();
 717
 718                 $ns = NS_MAIN;
 719
 720                 if ( $disableLinkConversion ||
 721                          ( !$ignoreOtherCond &&
 722                            ( $isredir == 'no'
 723                                  || $action == 'edit'
 724                                  || $action == 'submit'
 725                                  || $linkconvert == 'no'
 726                                  || $wgUser->getOption( 'noconvertlink' ) == 1 ) ) ) {
 727                         return;
 728                 }
 729
 730                 if ( is_object( $nt ) ) {
 731                         $ns = $nt->getNamespace();
 732                 }
 733
 734                 $variants = $this->autoConvertToAllVariants( $link );
 735                 if ( !$variants ) { // give up
 736                         return;
 737                 }
 738
 739                 $titles = array();
 740
 741                 foreach ( $variants as $v ) {
 742                         if ( $v != $link ) {
 743                                 $varnt = Title::newFromText( $v, $ns );
 744                                 if ( !is_null( $varnt ) ) {
 745                                         $linkBatch->addObj( $varnt );
 746                                         $titles[] = $varnt;
 747                                 }
 748                         }
 749                 }
 750
 751                 // fetch all variants in single query
 752                 $linkBatch->execute();
 753
 754                 foreach ( $titles as $varnt ) {
 755                         if ( $varnt->getArticleID() > 0 ) {
 756                                 $nt = $varnt;
 757                                 $link = $varnt->getText();
 758                                 break;
 759                         }
 760                 }
 761         }
 762
 763         /**
 764          * Returns language specific hash options.
 765          */
 766         public function getExtraHashOptions() {
 767                 $variant = $this->getPreferredVariant();
 768                 return '!' . $variant;
 769         }
 770
 771         /**
 772          * Load default conversion tables.
 773          * This method must be implemented in derived class.
 774          *
 775          * @private
 776          */
 777         function loadDefaultTables() {
 778                 $name = get_class( $this );
 779                 wfDie( "Must implement loadDefaultTables() method in class $name" );
 780         }
 781
 782         /**
 783          * Load conversion tables either from the cache or the disk.
 784          * @private
 785          * @param $fromCache Boolean: load from memcached? Defaults to true.
 786          */
 787         function loadTables( $fromCache = true ) {
 788                 if ( $this->mTablesLoaded ) {
 789                         return;
 790                 }
 791                 global $wgMemc;
 792                 wfProfileIn( __METHOD__ );
 793                 $this->mTablesLoaded = true;
 794                 $this->mTables = false;
 795                 if ( $fromCache ) {
 796                         wfProfileIn( __METHOD__ . '-cache' );
 797                         $this->mTables = $wgMemc->get( $this->mCacheKey );
 798                         wfProfileOut( __METHOD__ . '-cache' );
 799                 }
 800                 if ( !$this->mTables
 801                          || !array_key_exists( self::CACHE_VERSION_KEY, $this->mTables ) ) {
 802                         wfProfileIn( __METHOD__ . '-recache' );
 803                         // not in cache, or we need a fresh reload.
 804                         // We will first load the default tables
 805                         // then update them using things in MediaWiki:Conversiontable/*
 806                         $this->loadDefaultTables();
 807                         foreach ( $this->mVariants as $var ) {
 808                                 $cached = $this->parseCachedTable( $var );
 809                                 $this->mTables[$var]->mergeArray( $cached );
 810                         }
 811
 812                         $this->postLoadTables();
 813                         $this->mTables[self::CACHE_VERSION_KEY] = true;
 814
 815                         $wgMemc->set( $this->mCacheKey, $this->mTables, 43200 );
 816                         wfProfileOut( __METHOD__ . '-recache' );
 817                 }
 818                 wfProfileOut( __METHOD__ );
 819         }
 820
 821         /**
 822          * Hook for post processing after conversion tables are loaded.
 823          */
 824         function postLoadTables() { }
 825
 826         /**
 827          * Reload the conversion tables.
 828          *
 829          * @private
 830          */
 831         function reloadTables() {
 832                 if ( $this->mTables ) {
 833                         unset( $this->mTables );
 834                 }
 835                 $this->mTablesLoaded = false;
 836                 $this->loadTables( false );
 837         }
 838
 839         /**
 840          * Parse the conversion table stored in the cache.
 841          *
 842          * The tables should be in blocks of the following form:
 843          *              -{
 844          *                      word => word ;
 845          *                      word => word ;
 846          *                      ...
 847          *              }-
 848          *
 849          * To make the tables more manageable, subpages are allowed
 850          * and will be parsed recursively if $recursive == true.
 851          *
 852          * @param $code String: language code
 853          * @param $subpage String: subpage name
 854          * @param $recursive Boolean: parse subpages recursively? Defaults to true.
 855          */
 856         function parseCachedTable( $code, $subpage = '', $recursive = true ) {
 857                 static $parsed = array();
 858
 859                 $key = 'Conversiontable/' . $code;
 860                 if ( $subpage ) {
 861                         $key .= '/' . $subpage;
 862                 }
 863                 if ( array_key_exists( $key, $parsed ) ) {
 864                         return array();
 865                 }
 866
 867                 if ( strpos( $code, '/' ) === false ) {
 868                         $txt = MessageCache::singleton()->get( 'Conversiontable', true, $code );
 869                         if ( $txt === false ) {
 870                                 # FIXME: this method doesn't seem to be expecting
 871                                 # this possible outcome...
 872                                 $txt = '&lt;Conversiontable&gt;';
 873                         }
 874                 } else {
 875                         $title = Title::makeTitleSafe(
 876                                 NS_MEDIAWIKI,
 877                                 "Conversiontable/$code"
 878                         );
 879                         if ( $title && $title->exists() ) {
 880                                 $article = new Article( $title );
 881                                 $txt = $article->getContents();
 882                         } else {
 883                                 $txt = '';
 884                         }
 885                 }
 886
 887                 // get all subpage links of the form
 888                 // [[MediaWiki:Conversiontable/zh-xx/...|...]]
 889                 $linkhead = $this->mLangObj->getNsText( NS_MEDIAWIKI ) .
 890                         ':Conversiontable';
 891                 $subs = StringUtils::explode( '[[', $txt );
 892                 $sublinks = array();
 893                 foreach ( $subs as $sub ) {
 894                         $link = explode( ']]', $sub, 2 );
 895                         if ( count( $link ) != 2 ) {
 896                                 continue;
 897                         }
 898                         $b = explode( '|', $link[0], 2 );
 899                         $b = explode( '/', trim( $b[0] ), 3 );
 900                         if ( count( $b ) == 3 ) {
 901                                 $sublink = $b[2];
 902                         } else {
 903                                 $sublink = '';
 904                         }
 905
 906                         if ( $b[0] == $linkhead && $b[1] == $code ) {
 907                                 $sublinks[] = $sublink;
 908                         }
 909                 }
 910
 911                 // parse the mappings in this page
 912                 $blocks = StringUtils::explode( '-{', $txt );
 913                 $ret = array();
 914                 $first = true;
 915                 foreach ( $blocks as $block ) {
 916                         if ( $first ) {
 917                                 // Skip the part before the first -{
 918                                 $first = false;
 919                                 continue;
 920                         }
 921                         $mappings = explode( '}-', $block, 2 );
 922                         $stripped = str_replace( array( "'", '"', '*', '#' ), '',
 923                                                                          $mappings[0] );
 924                         $table = StringUtils::explode( ';', $stripped );
 925                         foreach ( $table as $t ) {
 926                                 $m = explode( '=>', $t, 3 );
 927                                 if ( count( $m ) != 2 ) {
 928                                         continue;
 929                                 }
 930                                 // trim any trailling comments starting with '//'
 931                                 $tt = explode( '//', $m[1], 2 );
 932                                 $ret[trim( $m[0] )] = trim( $tt[0] );
 933                         }
 934                 }
 935                 $parsed[$key] = true;
 936
 937                 // recursively parse the subpages
 938                 if ( $recursive ) {
 939                         foreach ( $sublinks as $link ) {
 940                                 $s = $this->parseCachedTable( $code, $link, $recursive );
 941                                 $ret = array_merge( $ret, $s );
 942                         }
 943                 }
 944
 945                 if ( $this->mUcfirst ) {
 946                         foreach ( $ret as $k => $v ) {
 947                                 $ret[$this->mLangObj->ucfirst( $k )] = $this->mLangObj->ucfirst( $v );
 948                         }
 949                 }
 950                 return $ret;
 951         }
 952
 953         /**
 954          * Enclose a string with the "no conversion" tag. This is used by
 955          * various functions in the Parser.
 956          *
 957          * @param $text String: text to be tagged for no conversion
 958          * @param $noParse Boolean: unused
 959          * @return String: the tagged text
 960          */
 961         public function markNoConversion( $text, $noParse = false ) {
 962                 # don't mark if already marked
 963                 if ( strpos( $text, '-{' ) || strpos( $text, '}-' ) ) {
 964                         return $text;
 965                 }
 966
 967                 $ret = "-{R|$text}-";
 968                 return $ret;
 969         }
 970
 971         /**
 972          * Convert the sorting key for category links. This should make different
 973          * keys that are variants of each other map to the same key.
 974          */
 975         function convertCategoryKey( $key ) {
 976                 return $key;
 977         }
 978
 979         /**
 980          * Hook to refresh the cache of conversion tables when
 981          * MediaWiki:Conversiontable* is updated.
 982          * @private
 983          *
 984          * @param $article Object: Article object
 985          * @param $user Object: User object for the current user
 986          * @param $text String: article text (?)
 987          * @param $summary String: edit summary of the edit
 988          * @param $isMinor Boolean: was the edit marked as minor?
 989          * @param $isWatch Boolean: did the user watch this page or not?
 990          * @param $section Unused
 991          * @param $flags Bitfield
 992          * @param $revision Object: new Revision object or null
 993          * @return Boolean: true
 994          */
 995         function OnArticleSaveComplete( $article, $user, $text, $summary, $isMinor,
 996                         $isWatch, $section, $flags, $revision ) {
 997                 $titleobj = $article->getTitle();
 998                 if ( $titleobj->getNamespace() == NS_MEDIAWIKI ) {
 999                         $title = $titleobj->getDBkey();
1000                         $t = explode( '/', $title, 3 );
1001                         $c = count( $t );
1002                         if ( $c > 1 && $t[0] == 'Conversiontable' ) {
1003                                 if ( $this->validateVariant( $t[1] ) ) {
1004                                         $this->reloadTables();
1005                                 }
1006                         }
1007                 }
1008                 return true;
1009         }
1010
1011         /**
1012          * Armour rendered math against conversion.
1013          * Escape special chars in parsed math text. (in most cases are img elements)
1014          *
1015          * @param $text String: text to armour against conversion
1016          * @return String: armoured text where { and } have been converted to
1017          *                 &#123; and &#125;
1018          */
1019         public function armourMath( $text ) {
1020                 // convert '-{' and '}-' to '-&#123;' and '&#125;-' to prevent
1021                 // any unwanted markup appearing in the math image tag.
1022                 $text = strtr( $text, array( '-{' => '-&#123;', '}-' => '&#125;-' ) );
1023                 return $text;
1024         }
1025
1026         /**
1027          * Get the cached separator pattern for ConverterRule::parseRules()
1028          */
1029         function getVarSeparatorPattern() {
1030                 if ( is_null( $this->mVarSeparatorPattern ) ) {
1031                         // varsep_pattern for preg_split:
1032                         // text should be splited by ";" only if a valid variant
1033                         // name exist after the markup, for example:
1034                         //  -{zh-hans:<span style="font-size:120%;">xxx</span>;zh-hant:\
1035                         //      <span style="font-size:120%;">yyy</span>;}-
1036                         // we should split it as:
1037                         //  array(
1038                         //        [0] => 'zh-hans:<span style="font-size:120%;">xxx</span>'
1039                         //        [1] => 'zh-hant:<span style="font-size:120%;">yyy</span>'
1040                         //        [2] => ''
1041                         //       )
1042                         $pat = '/;\s*(?=';
1043                         foreach ( $this->mVariants as $variant ) {
1044                                 // zh-hans:xxx;zh-hant:yyy
1045                                 $pat .= $variant . '\s*:|';
1046                                 // xxx=>zh-hans:yyy; xxx=>zh-hant:zzz
1047                                 $pat .= '[^;]*?=>\s*' . $variant . '\s*:|';
1048                         }
1049                         $pat .= '\s*$)/';
1050                         $this->mVarSeparatorPattern = $pat;
1051                 }
1052                 return $this->mVarSeparatorPattern;
1053         }
1054 }
1055
1056 /**
1057  * Parser for rules of language conversion , parse rules in -{ }- tag.
1058  * @ingroup Language
1059  * @author fdcn <fdcn64@gmail.com>, PhiLiP <philip.npc@gmail.com>
1060  */
1061 class ConverterRule {
1062         var $mText; // original text in -{text}-
1063         var $mConverter; // LanguageConverter object
1064         var $mManualCodeError = '<strong class="error">code error!</strong>';
1065         var $mRuleDisplay = '';
1066         var $mRuleTitle = false;
1067         var $mRules = '';// string : the text of the rules
1068         var $mRulesAction = 'none';
1069         var $mFlags = array();
1070         var $mVariantFlags = array();
1071         var $mConvTable = array();
1072         var $mBidtable = array();// array of the translation in each variant
1073         var $mUnidtable = array();// array of the translation in each variant
1074
1075         /**
1076          * Constructor
1077          *
1078          * @param $text String: the text between -{ and }-
1079          * @param $converter LanguageConverter object
1080          */
1081         public function __construct( $text, $converter ) {
1082                 $this->mText = $text;
1083                 $this->mConverter = $converter;
1084         }
1085
1086         /**
1087          * Check if variants array in convert array.
1088          *
1089          * @param $variants Array or string: variant language code
1090          * @return String: translated text
1091          */
1092         public function getTextInBidtable( $variants ) {
1093                 $variants = (array)$variants;
1094                 if ( !$variants ) {
1095                         return false;
1096                 }
1097                 foreach ( $variants as $variant ) {
1098                         if ( isset( $this->mBidtable[$variant] ) ) {
1099                                 return $this->mBidtable[$variant];
1100                         }
1101                 }
1102                 return false;
1103         }
1104
1105         /**
1106          * Parse flags with syntax -{FLAG| ... }-
1107          * @private
1108          */
1109         function parseFlags() {
1110                 $text = $this->mText;
1111                 $flags = array();
1112                 $variantFlags = array();
1113
1114                 $sepPos = strpos( $text, '|' );
1115                 if ( $sepPos !== false ) {
1116                         $validFlags = $this->mConverter->mFlags;
1117                         $f = StringUtils::explode( ';', substr( $text, 0, $sepPos ) );
1118                         foreach ( $f as $ff ) {
1119                                 $ff = trim( $ff );
1120                                 if ( isset( $validFlags[$ff] ) ) {
1121                                         $flags[$validFlags[$ff]] = true;
1122                                 }
1123                         }
1124                         $text = strval( substr( $text, $sepPos + 1 ) );
1125                 }
1126
1127                 if ( !$flags ) {
1128                         $flags['S'] = true;
1129                 } elseif ( isset( $flags['R'] ) ) {
1130                         $flags = array( 'R' => true );// remove other flags
1131                 } elseif ( isset( $flags['N'] ) ) {
1132                         $flags = array( 'N' => true );// remove other flags
1133                 } elseif ( isset( $flags['-'] ) ) {
1134                         $flags = array( '-' => true );// remove other flags
1135                 } elseif ( count( $flags ) == 1 && isset( $flags['T'] ) ) {
1136                         $flags['H'] = true;
1137                 } elseif ( isset( $flags['H'] ) ) {
1138                         // replace A flag, and remove other flags except T
1139                         $temp = array( '+' => true, 'H' => true );
1140                         if ( isset( $flags['T'] ) ) {
1141                                 $temp['T'] = true;
1142                         }
1143                         if ( isset( $flags['D'] ) ) {
1144                                 $temp['D'] = true;
1145                         }
1146                         $flags = $temp;
1147                 } else {
1148                         if ( isset( $flags['A'] ) ) {
1149                                 $flags['+'] = true;
1150                                 $flags['S'] = true;
1151                         }
1152                         if ( isset( $flags['D'] ) ) {
1153                                 unset( $flags['S'] );
1154                         }
1155                         // try to find flags like "zh-hans", "zh-hant"
1156                         // allow syntaxes like "-{zh-hans;zh-hant|XXXX}-"
1157                         $variantFlags = array_intersect( array_keys( $flags ), $this->mConverter->mVariants );
1158                         if ( $variantFlags ) {
1159                                 $variantFlags = array_flip( $variantFlags );
1160                                 $flags = array();
1161                         }
1162                 }
1163                 $this->mVariantFlags = $variantFlags;
1164                 $this->mRules = $text;
1165                 $this->mFlags = $flags;
1166         }
1167
1168         /**
1169          * Generate conversion table.
1170          * @private
1171          */
1172         function parseRules() {
1173                 $rules = $this->mRules;
1174                 $bidtable = array();
1175                 $unidtable = array();
1176                 $variants = $this->mConverter->mVariants;
1177                 $varsep_pattern = $this->mConverter->getVarSeparatorPattern();
1178
1179                 $choice = preg_split( $varsep_pattern, $rules );
1180
1181                 foreach ( $choice as $c ) {
1182                         $v  = explode( ':', $c, 2 );
1183                         if ( count( $v ) != 2 ) {
1184                                 // syntax error, skip
1185                                 continue;
1186                         }
1187                         $to = trim( $v[1] );
1188                         $v  = trim( $v[0] );
1189                         $u  = explode( '=>', $v, 2 );
1190                         // if $to is empty, strtr() could return a wrong result
1191                         if ( count( $u ) == 1 && $to && in_array( $v, $variants ) ) {
1192                                 $bidtable[$v] = $to;
1193                         } elseif ( count( $u ) == 2 ) {
1194                                 $from = trim( $u[0] );
1195                                 $v      = trim( $u[1] );
1196                                 if ( array_key_exists( $v, $unidtable )
1197                                          && !is_array( $unidtable[$v] )
1198                                          && $to
1199                                          && in_array( $v, $variants ) ) {
1200                                         $unidtable[$v] = array( $from => $to );
1201                                 } elseif ( $to && in_array( $v, $variants ) ) {
1202                                         $unidtable[$v][$from] = $to;
1203                                 }
1204                         }
1205                         // syntax error, pass
1206                         if ( !isset( $this->mConverter->mVariantNames[$v] ) ) {
1207                                 $bidtable = array();
1208                                 $unidtable = array();
1209                                 break;
1210                         }
1211                 }
1212                 $this->mBidtable = $bidtable;
1213                 $this->mUnidtable = $unidtable;
1214         }
1215
1216         /**
1217          * @private
1218          */
1219         function getRulesDesc() {
1220                 $codesep = $this->mConverter->mDescCodeSep;
1221                 $varsep = $this->mConverter->mDescVarSep;
1222                 $text = '';
1223                 foreach ( $this->mBidtable as $k => $v ) {
1224                         $text .= $this->mConverter->mVariantNames[$k] . "$codesep$v$varsep";
1225                 }
1226                 foreach ( $this->mUnidtable as $k => $a ) {
1227                         foreach ( $a as $from => $to ) {
1228                                 $text .= $from . '⇒' . $this->mConverter->mVariantNames[$k] .
1229                                         "$codesep$to$varsep";
1230                         }
1231                 }
1232                 return $text;
1233         }
1234
1235         /**
1236          * Parse rules conversion.
1237          * @private
1238          */
1239         function getRuleConvertedStr( $variant ) {
1240                 $bidtable = $this->mBidtable;
1241                 $unidtable = $this->mUnidtable;
1242
1243                 if ( count( $bidtable ) + count( $unidtable ) == 0 ) {
1244                         return $this->mRules;
1245                 } else {
1246                         // display current variant in bidirectional array
1247                         $disp = $this->getTextInBidtable( $variant );
1248                         // or display current variant in fallbacks
1249                         if ( !$disp ) {
1250                                 $disp = $this->getTextInBidtable(
1251                                                 $this->mConverter->getVariantFallbacks( $variant ) );
1252                         }
1253                         // or display current variant in unidirectional array
1254                         if ( !$disp && array_key_exists( $variant, $unidtable ) ) {
1255                                 $disp = array_values( $unidtable[$variant] );
1256                                 $disp = $disp[0];
1257                         }
1258                         // or display frist text under disable manual convert
1259                         if ( !$disp
1260                                  && $this->mConverter->mManualLevel[$variant] == 'disable' ) {
1261                                 if ( count( $bidtable ) > 0 ) {
1262                                         $disp = array_values( $bidtable );
1263                                         $disp = $disp[0];
1264                                 } else {
1265                                         $disp = array_values( $unidtable );
1266                                         $disp = array_values( $disp[0] );
1267                                         $disp = $disp[0];
1268                                 }
1269                         }
1270                         return $disp;
1271                 }
1272         }
1273
1274         /**
1275          * Generate conversion table for all text.
1276          * @private
1277          */
1278         function generateConvTable() {
1279                 // Special case optimisation
1280                 if ( !$this->mBidtable && !$this->mUnidtable ) {
1281                         $this->mConvTable = array();
1282                         return;
1283                 }
1284
1285                 $bidtable = $this->mBidtable;
1286                 $unidtable = $this->mUnidtable;
1287                 $manLevel = $this->mConverter->mManualLevel;
1288
1289                 $vmarked = array();
1290                 foreach ( $this->mConverter->mVariants as $v ) {
1291                         /* for bidirectional array
1292                                 fill in the missing variants, if any,
1293                                 with fallbacks */
1294                         if ( !isset( $bidtable[$v] ) ) {
1295                                 $variantFallbacks =
1296                                         $this->mConverter->getVariantFallbacks( $v );
1297                                 $vf = $this->getTextInBidtable( $variantFallbacks );
1298                                 if ( $vf ) {
1299                                         $bidtable[$v] = $vf;
1300                                 }
1301                         }
1302
1303                         if ( isset( $bidtable[$v] ) ) {
1304                                 foreach ( $vmarked as $vo ) {
1305                                         // use syntax: -{A|zh:WordZh;zh-tw:WordTw}-
1306                                         // or -{H|zh:WordZh;zh-tw:WordTw}-
1307                                         // or -{-|zh:WordZh;zh-tw:WordTw}-
1308                                         // to introduce a custom mapping between
1309                                         // words WordZh and WordTw in the whole text
1310                                         if ( $manLevel[$v] == 'bidirectional' ) {
1311                                                 $this->mConvTable[$v][$bidtable[$vo]] = $bidtable[$v];
1312                                         }
1313                                         if ( $manLevel[$vo] == 'bidirectional' ) {
1314                                                 $this->mConvTable[$vo][$bidtable[$v]] = $bidtable[$vo];
1315                                         }
1316                                 }
1317                                 $vmarked[] = $v;
1318                         }
1319                         /* for unidirectional array fill to convert tables */
1320                         if ( ( $manLevel[$v] == 'bidirectional' || $manLevel[$v] == 'unidirectional' )
1321                                 && isset( $unidtable[$v] ) )
1322                         {
1323                                 if ( isset( $this->mConvTable[$v] ) ) {
1324                                         $this->mConvTable[$v] = array_merge( $this->mConvTable[$v], $unidtable[$v] );
1325                                 } else {
1326                                         $this->mConvTable[$v] = $unidtable[$v];
1327                                 }
1328                         }
1329                 }
1330         }
1331
1332         /**
1333          * Parse rules and flags.
1334          * @param $variant String: variant language code
1335          */
1336         public function parse( $variant = null ) {
1337                 if ( !$variant ) {
1338                         $variant = $this->mConverter->getPreferredVariant();
1339                 }
1340
1341                 $this->parseFlags();
1342                 $flags = $this->mFlags;
1343
1344                 // convert to specified variant
1345                 // syntax: -{zh-hans;zh-hant[;...]|<text to convert>}-
1346                 if ( $this->mVariantFlags ) {
1347                         // check if current variant in flags
1348                         if ( isset( $this->mVariantFlags[$variant] ) ) {
1349                                 // then convert <text to convert> to current language
1350                                 $this->mRules = $this->mConverter->autoConvert( $this->mRules,
1351                                                                                                                                 $variant );
1352                         } else { // if current variant no in flags,
1353                                    // then we check its fallback variants.
1354                                 $variantFallbacks =
1355                                         $this->mConverter->getVariantFallbacks( $variant );
1356                                 foreach ( $variantFallbacks as $variantFallback ) {
1357                                         // if current variant's fallback exist in flags
1358                                         if ( isset( $this->mVariantFlags[$variantFallback] ) ) {
1359                                                 // then convert <text to convert> to fallback language
1360                                                 $this->mRules =
1361                                                         $this->mConverter->autoConvert( $this->mRules,
1362                                                                                                                         $variantFallback );
1363                                                 break;
1364                                         }
1365                                 }
1366                         }
1367                         $this->mFlags = $flags = array( 'R' => true );
1368                 }
1369
1370                 if ( !isset( $flags['R'] ) && !isset( $flags['N'] ) ) {
1371                         // decode => HTML entities modified by Sanitizer::removeHTMLtags
1372                         $this->mRules = str_replace( '=&gt;', '=>', $this->mRules );
1373                         $this->parseRules();
1374                 }
1375                 $rules = $this->mRules;
1376
1377                 if ( !$this->mBidtable && !$this->mUnidtable ) {
1378                         if ( isset( $flags['+'] ) || isset( $flags['-'] ) ) {
1379                                 // fill all variants if text in -{A/H/-|text} without rules
1380                                 foreach ( $this->mConverter->mVariants as $v ) {
1381                                         $this->mBidtable[$v] = $rules;
1382                                 }
1383                         } elseif ( !isset( $flags['N'] ) && !isset( $flags['T'] ) ) {
1384                                 $this->mFlags = $flags = array( 'R' => true );
1385                         }
1386                 }
1387
1388                 $this->mRuleDisplay = false;
1389                 foreach ( $flags as $flag => $unused ) {
1390                         switch ( $flag ) {
1391                                 case 'R':
1392                                         // if we don't do content convert, still strip the -{}- tags
1393                                         $this->mRuleDisplay = $rules;
1394                                         break;
1395                                 case 'N':
1396                                         // process N flag: output current variant name
1397                                         $ruleVar = trim( $rules );
1398                                         if ( isset( $this->mConverter->mVariantNames[$ruleVar] ) ) {
1399                                                 $this->mRuleDisplay = $this->mConverter->mVariantNames[$ruleVar];
1400                                         } else {
1401                                                 $this->mRuleDisplay = '';
1402                                         }
1403                                         break;
1404                                 case 'D':
1405                                         // process D flag: output rules description
1406                                         $this->mRuleDisplay = $this->getRulesDesc();
1407                                         break;
1408                                 case 'H':
1409                                         // process H,- flag or T only: output nothing
1410                                         $this->mRuleDisplay = '';
1411                                         break;
1412                                 case '-':
1413                                         $this->mRulesAction = 'remove';
1414                                         $this->mRuleDisplay = '';
1415                                         break;
1416                                 case '+':
1417                                         $this->mRulesAction = 'add';
1418                                         $this->mRuleDisplay = '';
1419                                         break;
1420                                 case 'S':
1421                                         $this->mRuleDisplay = $this->getRuleConvertedStr( $variant );
1422                                         break;
1423                                 case 'T':
1424                                         $this->mRuleTitle = $this->getRuleConvertedStr( $variant );
1425                                         $this->mRuleDisplay = '';
1426                                         break;
1427                                 default:
1428                                         // ignore unknown flags (but see error case below)
1429                         }
1430                 }
1431                 if ( $this->mRuleDisplay === false ) {
1432                         $this->mRuleDisplay = $this->mManualCodeError;
1433                 }
1434
1435                 $this->generateConvTable();
1436         }
1437
1438         /**
1439          * @todo FIXME: code this function :)
1440          */
1441         public function hasRules() {
1442                 // TODO:
1443         }
1444
1445         /**
1446          * Get display text on markup -{...}-
1447          */
1448         public function getDisplay() {
1449                 return $this->mRuleDisplay;
1450         }
1451
1452         /**
1453          * Get converted title.
1454          */
1455         public function getTitle() {
1456                 return $this->mRuleTitle;
1457         }
1458
1459         /**
1460          * Return how deal with conversion rules.
1461          */
1462         public function getRulesAction() {
1463                 return $this->mRulesAction;
1464         }
1465
1466         /**
1467          * Get conversion table. (bidirectional and unidirectional
1468          * conversion table)
1469          */
1470         public function getConvTable() {
1471                 return $this->mConvTable;
1472         }
1473
1474         /**
1475          * Get conversion rules string.
1476          */
1477         public function getRules() {
1478                 return $this->mRules;
1479         }
1480
1481         /**
1482          * Get conversion flags.
1483          */
1484         public function getFlags() {
1485                 return $this->mFlags;
1486         }
1487 }