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