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::fetchLanguageNames();
  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 ) {
 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                 $shouldConvert = !$this->guessVariant( $text, $variant );
 599
 600                 while ( $startPos < $length ) {
 601                         $pos = strpos( $text, '-{', $startPos );
 602
 603                         if ( $pos === false ) {
 604                                 // No more markup, append final segment
 605                                 $fragment = substr( $text, $startPos );
 606                                 $out .= $shouldConvert? $this->autoConvert( $fragment, $variant ): $fragment;
 607                                 return $out;
 608                         }
 609
 610                         // Markup found
 611                         // Append initial segment
 612                         $fragment = substr( $text, $startPos, $pos - $startPos );
 613                         $out .= $shouldConvert? $this->autoConvert( $fragment, $variant ): $fragment;
 614
 615                         // Advance position
 616                         $startPos = $pos;
 617
 618                         // Do recursive conversion
 619                         $out .= $this->recursiveConvertRule( $text, $variant, $startPos, $depth + 1 );
 620                 }
 621
 622                 return $out;
 623         }
 624
 625         /**
 626          * Recursively convert text on the inside.
 627          *
 628          * @param $text String: text to be converted
 629          * @param $variant String: the target variant code
 630          * @param $startPos int
 631          * @param $depth Integer: depth of recursion
 632          *
 633          * @throws MWException
 634          * @return String: converted text
 635          */
 636         protected function recursiveConvertRule( $text, $variant, &$startPos, $depth = 0 ) {
 637                 // Quick sanity check (no function calls)
 638                 if ( $text[$startPos] !== '-' || $text[$startPos + 1] !== '{' ) {
 639                         throw new MWException( __METHOD__ . ': invalid input string' );
 640                 }
 641
 642                 $startPos += 2;
 643                 $inner = '';
 644                 $warningDone = false;
 645                 $length = strlen( $text );
 646
 647                 while ( $startPos < $length ) {
 648                         $m = false;
 649                         preg_match( '/-\{|\}-/', $text, $m,  PREG_OFFSET_CAPTURE, $startPos );
 650                         if ( !$m ) {
 651                                 // Unclosed rule
 652                                 break;
 653                         }
 654
 655                         $token = $m[0][0];
 656                         $pos = $m[0][1];
 657
 658                         // Markup found
 659                         // Append initial segment
 660                         $inner .= substr( $text, $startPos, $pos - $startPos );
 661
 662                         // Advance position
 663                         $startPos = $pos;
 664
 665                         switch ( $token ) {
 666                                 case '-{':
 667                                         // Check max depth
 668                                         if ( $depth >= $this->mMaxDepth ) {
 669                                                 $inner .= '-{';
 670                                                 if ( !$warningDone ) {
 671                                                         $inner .= '<span class="error">' .
 672                                                                 wfMsgForContent( 'language-converter-depth-warning',
 673                                                                         $this->mMaxDepth ) .
 674                                                                 '</span>';
 675                                                         $warningDone = true;
 676                                                 }
 677                                                 $startPos += 2;
 678                                                 continue;
 679                                         }
 680                                         // Recursively parse another rule
 681                                         $inner .= $this->recursiveConvertRule( $text, $variant, $startPos, $depth + 1 );
 682                                         break;
 683                                 case '}-':
 684                                         // Apply the rule
 685                                         $startPos += 2;
 686                                         $rule = new ConverterRule( $inner, $this );
 687                                         $rule->parse( $variant );
 688                                         $this->applyManualConv( $rule );
 689                                         return $rule->getDisplay();
 690                                 default:
 691                                         throw new MWException( __METHOD__ . ': invalid regex match' );
 692                         }
 693                 }
 694
 695                 // Unclosed rule
 696                 if ( $startPos < $length ) {
 697                         $inner .= substr( $text, $startPos );
 698                 }
 699                 $startPos = $length;
 700                 return '-{' . $this->autoConvert( $inner, $variant );
 701         }
 702
 703         /**
 704          * If a language supports multiple variants, it is possible that
 705          * non-existing link in one variant actually exists in another variant.
 706          * This function tries to find it. See e.g. LanguageZh.php
 707          *
 708          * @param $link String: the name of the link
 709          * @param $nt Mixed: the title object of the link
 710          * @param $ignoreOtherCond Boolean: to disable other conditions when
 711          *              we need to transclude a template or update a category's link
 712          * @return Null, the input parameters may be modified upon return
 713          */
 714         public function findVariantLink( &$link, &$nt, $ignoreOtherCond = false ) {
 715                 # If the article has already existed, there is no need to
 716                 # check it again, otherwise it may cause a fault.
 717                 if ( is_object( $nt ) && $nt->exists() ) {
 718                         return;
 719                 }
 720
 721                 global $wgDisableLangConversion, $wgDisableTitleConversion, $wgRequest,
 722                         $wgUser;
 723                 $isredir = $wgRequest->getText( 'redirect', 'yes' );
 724                 $action = $wgRequest->getText( 'action' );
 725                 $linkconvert = $wgRequest->getText( 'linkconvert', 'yes' );
 726                 $disableLinkConversion = $wgDisableLangConversion
 727                         || $wgDisableTitleConversion;
 728                 $linkBatch = new LinkBatch();
 729
 730                 $ns = NS_MAIN;
 731
 732                 if ( $disableLinkConversion ||
 733                          ( !$ignoreOtherCond &&
 734                            ( $isredir == 'no'
 735                                  || $action == 'edit'
 736                                  || $action == 'submit'
 737                                  || $linkconvert == 'no'
 738                                  || $wgUser->getOption( 'noconvertlink' ) == 1 ) ) ) {
 739                         return;
 740                 }
 741
 742                 if ( is_object( $nt ) ) {
 743                         $ns = $nt->getNamespace();
 744                 }
 745
 746                 $variants = $this->autoConvertToAllVariants( $link );
 747                 if ( !$variants ) { // give up
 748                         return;
 749                 }
 750
 751                 $titles = array();
 752
 753                 foreach ( $variants as $v ) {
 754                         if ( $v != $link ) {
 755                                 $varnt = Title::newFromText( $v, $ns );
 756                                 if ( !is_null( $varnt ) ) {
 757                                         $linkBatch->addObj( $varnt );
 758                                         $titles[] = $varnt;
 759                                 }
 760                         }
 761                 }
 762
 763                 // fetch all variants in single query
 764                 $linkBatch->execute();
 765
 766                 foreach ( $titles as $varnt ) {
 767                         if ( $varnt->getArticleID() > 0 ) {
 768                                 $nt = $varnt;
 769                                 $link = $varnt->getText();
 770                                 break;
 771                         }
 772                 }
 773         }
 774
 775         /**
 776          * Returns language specific hash options.
 777          *
 778          * @return string
 779          */
 780         public function getExtraHashOptions() {
 781                 $variant = $this->getPreferredVariant();
 782                 return '!' . $variant;
 783         }
 784
 785         /**
 786          * Guess if a text is written in a variant. This should be implemented in subclasses.
 787          *
 788          * @param string        $text the text to be checked
 789          * @param string        $variant language code of the variant to be checked for
 790          * @return bool true if $text appears to be written in $variant, false if not
 791          *
 792          * @author Nikola Smolenski <smolensk@eunet.rs>
 793          * @since 1.19
 794          */
 795         public function guessVariant($text, $variant) {
 796                 return false;
 797         }
 798
 799         /**
 800          * Load default conversion tables.
 801          * This method must be implemented in derived class.
 802          *
 803          * @private
 804          * @throws MWException
 805          */
 806         function loadDefaultTables() {
 807                 $name = get_class( $this );
 808                 throw new MWException( "Must implement loadDefaultTables() method in class $name" );
 809         }
 810
 811         /**
 812          * Load conversion tables either from the cache or the disk.
 813          * @private
 814          * @param $fromCache Boolean: load from memcached? Defaults to true.
 815          */
 816         function loadTables( $fromCache = true ) {
 817                 global $wgLangConvMemc;
 818
 819                 if ( $this->mTablesLoaded ) {
 820                         return;
 821                 }
 822
 823                 wfProfileIn( __METHOD__ );
 824                 $this->mTablesLoaded = true;
 825                 $this->mTables = false;
 826                 if ( $fromCache ) {
 827                         wfProfileIn( __METHOD__ . '-cache' );
 828                         $this->mTables = $wgLangConvMemc->get( $this->mCacheKey );
 829                         wfProfileOut( __METHOD__ . '-cache' );
 830                 }
 831                 if ( !$this->mTables
 832                          || !array_key_exists( self::CACHE_VERSION_KEY, $this->mTables ) ) {
 833                         wfProfileIn( __METHOD__ . '-recache' );
 834                         // not in cache, or we need a fresh reload.
 835                         // We will first load the default tables
 836                         // then update them using things in MediaWiki:Conversiontable/*
 837                         $this->loadDefaultTables();
 838                         foreach ( $this->mVariants as $var ) {
 839                                 $cached = $this->parseCachedTable( $var );
 840                                 $this->mTables[$var]->mergeArray( $cached );
 841                         }
 842
 843                         $this->postLoadTables();
 844                         $this->mTables[self::CACHE_VERSION_KEY] = true;
 845
 846                         $wgLangConvMemc->set( $this->mCacheKey, $this->mTables, 43200 );
 847                         wfProfileOut( __METHOD__ . '-recache' );
 848                 }
 849                 wfProfileOut( __METHOD__ );
 850         }
 851
 852         /**
 853          * Hook for post processing after conversion tables are loaded.
 854          */
 855         function postLoadTables() { }
 856
 857         /**
 858          * Reload the conversion tables.
 859          *
 860          * @private
 861          */
 862         function reloadTables() {
 863                 if ( $this->mTables ) {
 864                         unset( $this->mTables );
 865                 }
 866                 $this->mTablesLoaded = false;
 867                 $this->loadTables( false );
 868         }
 869
 870         /**
 871          * Parse the conversion table stored in the cache.
 872          *
 873          * The tables should be in blocks of the following form:
 874          *              -{
 875          *                      word => word ;
 876          *                      word => word ;
 877          *                      ...
 878          *              }-
 879          *
 880          * To make the tables more manageable, subpages are allowed
 881          * and will be parsed recursively if $recursive == true.
 882          *
 883          * @param $code String: language code
 884          * @param $subpage String: subpage name
 885          * @param $recursive Boolean: parse subpages recursively? Defaults to true.
 886          *
 887          * @return array
 888          */
 889         function parseCachedTable( $code, $subpage = '', $recursive = true ) {
 890                 static $parsed = array();
 891
 892                 $key = 'Conversiontable/' . $code;
 893                 if ( $subpage ) {
 894                         $key .= '/' . $subpage;
 895                 }
 896                 if ( array_key_exists( $key, $parsed ) ) {
 897                         return array();
 898                 }
 899
 900                 $parsed[$key] = true;
 901
 902                 if ( $subpage === '' ) {
 903                         $txt = MessageCache::singleton()->get( 'conversiontable', true, $code );
 904                 } else {
 905                         $txt = false;
 906                         $title = Title::makeTitleSafe( NS_MEDIAWIKI, $key );
 907                         if ( $title && $title->exists() ) {
 908                                 $revision = Revision::newFromTitle( $title );
 909                                 if ( $revision ) {
 910                                         $txt = $revision->getRawText();
 911                                 }
 912                         }
 913                 }
 914
 915                 # Nothing to parse if there's no text
 916                 if ( $txt === false || $txt === null || $txt === '' ) {
 917                         return array();
 918                 }
 919
 920                 // get all subpage links of the form
 921                 // [[MediaWiki:Conversiontable/zh-xx/...|...]]
 922                 $linkhead = $this->mLangObj->getNsText( NS_MEDIAWIKI ) .
 923                         ':Conversiontable';
 924                 $subs = StringUtils::explode( '[[', $txt );
 925                 $sublinks = array();
 926                 foreach ( $subs as $sub ) {
 927                         $link = explode( ']]', $sub, 2 );
 928                         if ( count( $link ) != 2 ) {
 929                                 continue;
 930                         }
 931                         $b = explode( '|', $link[0], 2 );
 932                         $b = explode( '/', trim( $b[0] ), 3 );
 933                         if ( count( $b ) == 3 ) {
 934                                 $sublink = $b[2];
 935                         } else {
 936                                 $sublink = '';
 937                         }
 938
 939                         if ( $b[0] == $linkhead && $b[1] == $code ) {
 940                                 $sublinks[] = $sublink;
 941                         }
 942                 }
 943
 944                 // parse the mappings in this page
 945                 $blocks = StringUtils::explode( '-{', $txt );
 946                 $ret = array();
 947                 $first = true;
 948                 foreach ( $blocks as $block ) {
 949                         if ( $first ) {
 950                                 // Skip the part before the first -{
 951                                 $first = false;
 952                                 continue;
 953                         }
 954                         $mappings = explode( '}-', $block, 2 );
 955                         $stripped = str_replace( array( "'", '"', '*', '#' ), '',
 956                                                                          $mappings[0] );
 957                         $table = StringUtils::explode( ';', $stripped );
 958                         foreach ( $table as $t ) {
 959                                 $m = explode( '=>', $t, 3 );
 960                                 if ( count( $m ) != 2 ) {
 961                                         continue;
 962                                 }
 963                                 // trim any trailling comments starting with '//'
 964                                 $tt = explode( '//', $m[1], 2 );
 965                                 $ret[trim( $m[0] )] = trim( $tt[0] );
 966                         }
 967                 }
 968
 969                 // recursively parse the subpages
 970                 if ( $recursive ) {
 971                         foreach ( $sublinks as $link ) {
 972                                 $s = $this->parseCachedTable( $code, $link, $recursive );
 973                                 $ret = array_merge( $ret, $s );
 974                         }
 975                 }
 976
 977                 if ( $this->mUcfirst ) {
 978                         foreach ( $ret as $k => $v ) {
 979                                 $ret[$this->mLangObj->ucfirst( $k )] = $this->mLangObj->ucfirst( $v );
 980                         }
 981                 }
 982                 return $ret;
 983         }
 984
 985         /**
 986          * Enclose a string with the "no conversion" tag. This is used by
 987          * various functions in the Parser.
 988          *
 989          * @param $text String: text to be tagged for no conversion
 990          * @param $noParse Boolean: unused
 991          * @return String: the tagged text
 992          */
 993         public function markNoConversion( $text, $noParse = false ) {
 994                 # don't mark if already marked
 995                 if ( strpos( $text, '-{' ) || strpos( $text, '}-' ) ) {
 996                         return $text;
 997                 }
 998
 999                 $ret = "-{R|$text}-";
1000                 return $ret;
1001         }
1002
1003         /**
1004          * Convert the sorting key for category links. This should make different
1005          * keys that are variants of each other map to the same key.
1006          *
1007          * @param $key string
1008          *
1009          * @return string
1010          */
1011         function convertCategoryKey( $key ) {
1012                 return $key;
1013         }
1014
1015         /**
1016          * Hook to refresh the cache of conversion tables when
1017          * MediaWiki:Conversiontable* is updated.
1018          * @private
1019          *
1020          * @param $article Article object
1021          * @param $user Object: User object for the current user
1022          * @param $text String: article text (?)
1023          * @param $summary String: edit summary of the edit
1024          * @param $isMinor Boolean: was the edit marked as minor?
1025          * @param $isWatch Boolean: did the user watch this page or not?
1026          * @param $section
1027          * @param $flags int Bitfield
1028          * @param $revision Object: new Revision object or null
1029          * @return Boolean: true
1030          */
1031         function OnArticleSaveComplete( $article, $user, $text, $summary, $isMinor,
1032                         $isWatch, $section, $flags, $revision ) {
1033                 $titleobj = $article->getTitle();
1034                 if ( $titleobj->getNamespace() == NS_MEDIAWIKI ) {
1035                         $title = $titleobj->getDBkey();
1036                         $t = explode( '/', $title, 3 );
1037                         $c = count( $t );
1038                         if ( $c > 1 && $t[0] == 'Conversiontable' ) {
1039                                 if ( $this->validateVariant( $t[1] ) ) {
1040                                         $this->reloadTables();
1041                                 }
1042                         }
1043                 }
1044                 return true;
1045         }
1046
1047         /**
1048          * Armour rendered math against conversion.
1049          * Escape special chars in parsed math text. (in most cases are img elements)
1050          *
1051          * @param $text String: text to armour against conversion
1052          * @return String: armoured text where { and } have been converted to
1053          *                 &#123; and &#125;
1054          */
1055         public function armourMath( $text ) {
1056                 // convert '-{' and '}-' to '-&#123;' and '&#125;-' to prevent
1057                 // any unwanted markup appearing in the math image tag.
1058                 $text = strtr( $text, array( '-{' => '-&#123;', '}-' => '&#125;-' ) );
1059                 return $text;
1060         }
1061
1062         /**
1063          * Get the cached separator pattern for ConverterRule::parseRules()
1064          */
1065         function getVarSeparatorPattern() {
1066                 if ( is_null( $this->mVarSeparatorPattern ) ) {
1067                         // varsep_pattern for preg_split:
1068                         // text should be splited by ";" only if a valid variant
1069                         // name exist after the markup, for example:
1070                         //  -{zh-hans:<span style="font-size:120%;">xxx</span>;zh-hant:\
1071                         //      <span style="font-size:120%;">yyy</span>;}-
1072                         // we should split it as:
1073                         //  array(
1074                         //        [0] => 'zh-hans:<span style="font-size:120%;">xxx</span>'
1075                         //        [1] => 'zh-hant:<span style="font-size:120%;">yyy</span>'
1076                         //        [2] => ''
1077                         //       )
1078                         $pat = '/;\s*(?=';
1079                         foreach ( $this->mVariants as $variant ) {
1080                                 // zh-hans:xxx;zh-hant:yyy
1081                                 $pat .= $variant . '\s*:|';
1082                                 // xxx=>zh-hans:yyy; xxx=>zh-hant:zzz
1083                                 $pat .= '[^;]*?=>\s*' . $variant . '\s*:|';
1084                         }
1085                         $pat .= '\s*$)/';
1086                         $this->mVarSeparatorPattern = $pat;
1087                 }
1088                 return $this->mVarSeparatorPattern;
1089         }
1090 }
1091
1092 /**
1093  * Parser for rules of language conversion , parse rules in -{ }- tag.
1094  * @ingroup Language
1095  * @author fdcn <fdcn64@gmail.com>, PhiLiP <philip.npc@gmail.com>
1096  */
1097 class ConverterRule {
1098         var $mText; // original text in -{text}-
1099         var $mConverter; // LanguageConverter object
1100         var $mManualCodeError = '<strong class="error">code error!</strong>';
1101         var $mRuleDisplay = '';
1102         var $mRuleTitle = false;
1103         var $mRules = '';// string : the text of the rules
1104         var $mRulesAction = 'none';
1105         var $mFlags = array();
1106         var $mVariantFlags = array();
1107         var $mConvTable = array();
1108         var $mBidtable = array();// array of the translation in each variant
1109         var $mUnidtable = array();// array of the translation in each variant
1110
1111         /**
1112          * Constructor
1113          *
1114          * @param $text String: the text between -{ and }-
1115          * @param $converter LanguageConverter object
1116          */
1117         public function __construct( $text, $converter ) {
1118                 $this->mText = $text;
1119                 $this->mConverter = $converter;
1120         }
1121
1122         /**
1123          * Check if variants array in convert array.
1124          *
1125          * @param $variants Array or string: variant language code
1126          * @return String: translated text
1127          */
1128         public function getTextInBidtable( $variants ) {
1129                 $variants = (array)$variants;
1130                 if ( !$variants ) {
1131                         return false;
1132                 }
1133                 foreach ( $variants as $variant ) {
1134                         if ( isset( $this->mBidtable[$variant] ) ) {
1135                                 return $this->mBidtable[$variant];
1136                         }
1137                 }
1138                 return false;
1139         }
1140
1141         /**
1142          * Parse flags with syntax -{FLAG| ... }-
1143          * @private
1144          */
1145         function parseFlags() {
1146                 $text = $this->mText;
1147                 $flags = array();
1148                 $variantFlags = array();
1149
1150                 $sepPos = strpos( $text, '|' );
1151                 if ( $sepPos !== false ) {
1152                         $validFlags = $this->mConverter->mFlags;
1153                         $f = StringUtils::explode( ';', substr( $text, 0, $sepPos ) );
1154                         foreach ( $f as $ff ) {
1155                                 $ff = trim( $ff );
1156                                 if ( isset( $validFlags[$ff] ) ) {
1157                                         $flags[$validFlags[$ff]] = true;
1158                                 }
1159                         }
1160                         $text = strval( substr( $text, $sepPos + 1 ) );
1161                 }
1162
1163                 if ( !$flags ) {
1164                         $flags['S'] = true;
1165                 } elseif ( isset( $flags['R'] ) ) {
1166                         $flags = array( 'R' => true );// remove other flags
1167                 } elseif ( isset( $flags['N'] ) ) {
1168                         $flags = array( 'N' => true );// remove other flags
1169                 } elseif ( isset( $flags['-'] ) ) {
1170                         $flags = array( '-' => true );// remove other flags
1171                 } elseif ( count( $flags ) == 1 && isset( $flags['T'] ) ) {
1172                         $flags['H'] = true;
1173                 } elseif ( isset( $flags['H'] ) ) {
1174                         // replace A flag, and remove other flags except T
1175                         $temp = array( '+' => true, 'H' => true );
1176                         if ( isset( $flags['T'] ) ) {
1177                                 $temp['T'] = true;
1178                         }
1179                         if ( isset( $flags['D'] ) ) {
1180                                 $temp['D'] = true;
1181                         }
1182                         $flags = $temp;
1183                 } else {
1184                         if ( isset( $flags['A'] ) ) {
1185                                 $flags['+'] = true;
1186                                 $flags['S'] = true;
1187                         }
1188                         if ( isset( $flags['D'] ) ) {
1189                                 unset( $flags['S'] );
1190                         }
1191                         // try to find flags like "zh-hans", "zh-hant"
1192                         // allow syntaxes like "-{zh-hans;zh-hant|XXXX}-"
1193                         $variantFlags = array_intersect( array_keys( $flags ), $this->mConverter->mVariants );
1194                         if ( $variantFlags ) {
1195                                 $variantFlags = array_flip( $variantFlags );
1196                                 $flags = array();
1197                         }
1198                 }
1199                 $this->mVariantFlags = $variantFlags;
1200                 $this->mRules = $text;
1201                 $this->mFlags = $flags;
1202         }
1203
1204         /**
1205          * Generate conversion table.
1206          * @private
1207          */
1208         function parseRules() {
1209                 $rules = $this->mRules;
1210                 $bidtable = array();
1211                 $unidtable = array();
1212                 $variants = $this->mConverter->mVariants;
1213                 $varsep_pattern = $this->mConverter->getVarSeparatorPattern();
1214
1215                 $choice = preg_split( $varsep_pattern, $rules );
1216
1217                 foreach ( $choice as $c ) {
1218                         $v  = explode( ':', $c, 2 );
1219                         if ( count( $v ) != 2 ) {
1220                                 // syntax error, skip
1221                                 continue;
1222                         }
1223                         $to = trim( $v[1] );
1224                         $v  = trim( $v[0] );
1225                         $u  = explode( '=>', $v, 2 );
1226                         // if $to is empty, strtr() could return a wrong result
1227                         if ( count( $u ) == 1 && $to && in_array( $v, $variants ) ) {
1228                                 $bidtable[$v] = $to;
1229                         } elseif ( count( $u ) == 2 ) {
1230                                 $from = trim( $u[0] );
1231                                 $v      = trim( $u[1] );
1232                                 if ( array_key_exists( $v, $unidtable )
1233                                          && !is_array( $unidtable[$v] )
1234                                          && $to
1235                                          && in_array( $v, $variants ) ) {
1236                                         $unidtable[$v] = array( $from => $to );
1237                                 } elseif ( $to && in_array( $v, $variants ) ) {
1238                                         $unidtable[$v][$from] = $to;
1239                                 }
1240                         }
1241                         // syntax error, pass
1242                         if ( !isset( $this->mConverter->mVariantNames[$v] ) ) {
1243                                 $bidtable = array();
1244                                 $unidtable = array();
1245                                 break;
1246                         }
1247                 }
1248                 $this->mBidtable = $bidtable;
1249                 $this->mUnidtable = $unidtable;
1250         }
1251
1252         /**
1253          * @private
1254          *
1255          * @return string
1256          */
1257         function getRulesDesc() {
1258                 $codesep = $this->mConverter->mDescCodeSep;
1259                 $varsep = $this->mConverter->mDescVarSep;
1260                 $text = '';
1261                 foreach ( $this->mBidtable as $k => $v ) {
1262                         $text .= $this->mConverter->mVariantNames[$k] . "$codesep$v$varsep";
1263                 }
1264                 foreach ( $this->mUnidtable as $k => $a ) {
1265                         foreach ( $a as $from => $to ) {
1266                                 $text .= $from . '⇒' . $this->mConverter->mVariantNames[$k] .
1267                                         "$codesep$to$varsep";
1268                         }
1269                 }
1270                 return $text;
1271         }
1272
1273         /**
1274          * Parse rules conversion.
1275          * @private
1276          *
1277          * @param $variant
1278          *
1279          * @return string
1280          */
1281         function getRuleConvertedStr( $variant ) {
1282                 $bidtable = $this->mBidtable;
1283                 $unidtable = $this->mUnidtable;
1284
1285                 if ( count( $bidtable ) + count( $unidtable ) == 0 ) {
1286                         return $this->mRules;
1287                 } else {
1288                         // display current variant in bidirectional array
1289                         $disp = $this->getTextInBidtable( $variant );
1290                         // or display current variant in fallbacks
1291                         if ( !$disp ) {
1292                                 $disp = $this->getTextInBidtable(
1293                                                 $this->mConverter->getVariantFallbacks( $variant ) );
1294                         }
1295                         // or display current variant in unidirectional array
1296                         if ( !$disp && array_key_exists( $variant, $unidtable ) ) {
1297                                 $disp = array_values( $unidtable[$variant] );
1298                                 $disp = $disp[0];
1299                         }
1300                         // or display frist text under disable manual convert
1301                         if ( !$disp
1302                                  && $this->mConverter->mManualLevel[$variant] == 'disable' ) {
1303                                 if ( count( $bidtable ) > 0 ) {
1304                                         $disp = array_values( $bidtable );
1305                                         $disp = $disp[0];
1306                                 } else {
1307                                         $disp = array_values( $unidtable );
1308                                         $disp = array_values( $disp[0] );
1309                                         $disp = $disp[0];
1310                                 }
1311                         }
1312                         return $disp;
1313                 }
1314         }
1315
1316         /**
1317          * Generate conversion table for all text.
1318          * @private
1319          */
1320         function generateConvTable() {
1321                 // Special case optimisation
1322                 if ( !$this->mBidtable && !$this->mUnidtable ) {
1323                         $this->mConvTable = array();
1324                         return;
1325                 }
1326
1327                 $bidtable = $this->mBidtable;
1328                 $unidtable = $this->mUnidtable;
1329                 $manLevel = $this->mConverter->mManualLevel;
1330
1331                 $vmarked = array();
1332                 foreach ( $this->mConverter->mVariants as $v ) {
1333                         /* for bidirectional array
1334                                 fill in the missing variants, if any,
1335                                 with fallbacks */
1336                         if ( !isset( $bidtable[$v] ) ) {
1337                                 $variantFallbacks =
1338                                         $this->mConverter->getVariantFallbacks( $v );
1339                                 $vf = $this->getTextInBidtable( $variantFallbacks );
1340                                 if ( $vf ) {
1341                                         $bidtable[$v] = $vf;
1342                                 }
1343                         }
1344
1345                         if ( isset( $bidtable[$v] ) ) {
1346                                 foreach ( $vmarked as $vo ) {
1347                                         // use syntax: -{A|zh:WordZh;zh-tw:WordTw}-
1348                                         // or -{H|zh:WordZh;zh-tw:WordTw}-
1349                                         // or -{-|zh:WordZh;zh-tw:WordTw}-
1350                                         // to introduce a custom mapping between
1351                                         // words WordZh and WordTw in the whole text
1352                                         if ( $manLevel[$v] == 'bidirectional' ) {
1353                                                 $this->mConvTable[$v][$bidtable[$vo]] = $bidtable[$v];
1354                                         }
1355                                         if ( $manLevel[$vo] == 'bidirectional' ) {
1356                                                 $this->mConvTable[$vo][$bidtable[$v]] = $bidtable[$vo];
1357                                         }
1358                                 }
1359                                 $vmarked[] = $v;
1360                         }
1361                         /* for unidirectional array fill to convert tables */
1362                         if ( ( $manLevel[$v] == 'bidirectional' || $manLevel[$v] == 'unidirectional' )
1363                                 && isset( $unidtable[$v] ) )
1364                         {
1365                                 if ( isset( $this->mConvTable[$v] ) ) {
1366                                         $this->mConvTable[$v] = array_merge( $this->mConvTable[$v], $unidtable[$v] );
1367                                 } else {
1368                                         $this->mConvTable[$v] = $unidtable[$v];
1369                                 }
1370                         }
1371                 }
1372         }
1373
1374         /**
1375          * Parse rules and flags.
1376          * @param $variant String: variant language code
1377          */
1378         public function parse( $variant = null ) {
1379                 if ( !$variant ) {
1380                         $variant = $this->mConverter->getPreferredVariant();
1381                 }
1382
1383                 $this->parseFlags();
1384                 $flags = $this->mFlags;
1385
1386                 // convert to specified variant
1387                 // syntax: -{zh-hans;zh-hant[;...]|<text to convert>}-
1388                 if ( $this->mVariantFlags ) {
1389                         // check if current variant in flags
1390                         if ( isset( $this->mVariantFlags[$variant] ) ) {
1391                                 // then convert <text to convert> to current language
1392                                 $this->mRules = $this->mConverter->autoConvert( $this->mRules,
1393                                         $variant );
1394                         } else { // if current variant no in flags,
1395                                    // then we check its fallback variants.
1396                                 $variantFallbacks =
1397                                         $this->mConverter->getVariantFallbacks( $variant );
1398                                 if( is_array( $variantFallbacks ) ) {
1399                                         foreach ( $variantFallbacks as $variantFallback ) {
1400                                                 // if current variant's fallback exist in flags
1401                                                 if ( isset( $this->mVariantFlags[$variantFallback] ) ) {
1402                                                         // then convert <text to convert> to fallback language
1403                                                         $this->mRules =
1404                                                                 $this->mConverter->autoConvert( $this->mRules,
1405                                                                         $variantFallback );
1406                                                         break;
1407                                                 }
1408                                         }
1409                                 }
1410                         }
1411                         $this->mFlags = $flags = array( 'R' => true );
1412                 }
1413
1414                 if ( !isset( $flags['R'] ) && !isset( $flags['N'] ) ) {
1415                         // decode => HTML entities modified by Sanitizer::removeHTMLtags
1416                         $this->mRules = str_replace( '=&gt;', '=>', $this->mRules );
1417                         $this->parseRules();
1418                 }
1419                 $rules = $this->mRules;
1420
1421                 if ( !$this->mBidtable && !$this->mUnidtable ) {
1422                         if ( isset( $flags['+'] ) || isset( $flags['-'] ) ) {
1423                                 // fill all variants if text in -{A/H/-|text} without rules
1424                                 foreach ( $this->mConverter->mVariants as $v ) {
1425                                         $this->mBidtable[$v] = $rules;
1426                                 }
1427                         } elseif ( !isset( $flags['N'] ) && !isset( $flags['T'] ) ) {
1428                                 $this->mFlags = $flags = array( 'R' => true );
1429                         }
1430                 }
1431
1432                 $this->mRuleDisplay = false;
1433                 foreach ( $flags as $flag => $unused ) {
1434                         switch ( $flag ) {
1435                                 case 'R':
1436                                         // if we don't do content convert, still strip the -{}- tags
1437                                         $this->mRuleDisplay = $rules;
1438                                         break;
1439                                 case 'N':
1440                                         // process N flag: output current variant name
1441                                         $ruleVar = trim( $rules );
1442                                         if ( isset( $this->mConverter->mVariantNames[$ruleVar] ) ) {
1443                                                 $this->mRuleDisplay = $this->mConverter->mVariantNames[$ruleVar];
1444                                         } else {
1445                                                 $this->mRuleDisplay = '';
1446                                         }
1447                                         break;
1448                                 case 'D':
1449                                         // process D flag: output rules description
1450                                         $this->mRuleDisplay = $this->getRulesDesc();
1451                                         break;
1452                                 case 'H':
1453                                         // process H,- flag or T only: output nothing
1454                                         $this->mRuleDisplay = '';
1455                                         break;
1456                                 case '-':
1457                                         $this->mRulesAction = 'remove';
1458                                         $this->mRuleDisplay = '';
1459                                         break;
1460                                 case '+':
1461                                         $this->mRulesAction = 'add';
1462                                         $this->mRuleDisplay = '';
1463                                         break;
1464                                 case 'S':
1465                                         $this->mRuleDisplay = $this->getRuleConvertedStr( $variant );
1466                                         break;
1467                                 case 'T':
1468                                         $this->mRuleTitle = $this->getRuleConvertedStr( $variant );
1469                                         $this->mRuleDisplay = '';
1470                                         break;
1471                                 default:
1472                                         // ignore unknown flags (but see error case below)
1473                         }
1474                 }
1475                 if ( $this->mRuleDisplay === false ) {
1476                         $this->mRuleDisplay = $this->mManualCodeError;
1477                 }
1478
1479                 $this->generateConvTable();
1480         }
1481
1482         /**
1483          * @todo FIXME: code this function :)
1484          */
1485         public function hasRules() {
1486                 // TODO:
1487         }
1488
1489         /**
1490          * Get display text on markup -{...}-
1491          * @return string
1492          */
1493         public function getDisplay() {
1494                 return $this->mRuleDisplay;
1495         }
1496
1497         /**
1498          * Get converted title.
1499          * @return string
1500          */
1501         public function getTitle() {
1502                 return $this->mRuleTitle;
1503         }
1504
1505         /**
1506          * Return how deal with conversion rules.
1507          * @return string
1508          */
1509         public function getRulesAction() {
1510                 return $this->mRulesAction;
1511         }
1512
1513         /**
1514          * Get conversion table. (bidirectional and unidirectional
1515          * conversion table)
1516          * @return array
1517          */
1518         public function getConvTable() {
1519                 return $this->mConvTable;
1520         }
1521
1522         /**
1523          * Get conversion rules string.
1524          * @return string
1525          */
1526         public function getRules() {
1527                 return $this->mRules;
1528         }
1529
1530         /**
1531          * Get conversion flags.
1532          * @return array
1533          */
1534         public function getFlags() {
1535                 return $this->mFlags;
1536         }
1537 }