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
  33         /**
  34          * languages supporting variants
  35          * @since 1.20
  36          * @var array
  37          */
  38         static public $languagesWithVariants = array(
  39                 'gan',
  40                 'iu',
  41                 'kk',
  42                 'ku',
  43                 'shi',
  44                 'sr',
  45                 'tg',
  46                 'uz',
  47                 'zh',
  48         );
  49
  50         public $mMainLanguageCode;
  51         public $mVariants, $mVariantFallbacks, $mVariantNames;
  52         public $mTablesLoaded = false;
  53         public $mTables;
  54         // 'bidirectional' 'unidirectional' 'disable' for each variant
  55         public $mManualLevel;
  56
  57         /**
  58          * @var String: memcached key name
  59          */
  60         public $mCacheKey;
  61
  62         public $mLangObj;
  63         public $mFlags;
  64         public $mDescCodeSep = ':', $mDescVarSep = ';';
  65         public $mUcfirst = false;
  66         public $mConvRuleTitle = false;
  67         public $mURLVariant;
  68         public $mUserVariant;
  69         public $mHeaderVariant;
  70         public $mMaxDepth = 10;
  71         public $mVarSeparatorPattern;
  72
  73         const CACHE_VERSION_KEY = 'VERSION 7';
  74
  75         /**
  76          * Constructor
  77          *
  78          * @param $langobj Language: the Language Object
  79          * @param $maincode String: the main language code of this language
  80          * @param $variants Array: the supported variants of this language
  81          * @param $variantfallbacks Array: the fallback language of each variant
  82          * @param $flags Array: defining the custom strings that maps to the flags
  83          * @param $manualLevel Array: limit for supported variants
  84          */
  85         public function __construct( $langobj, $maincode, $variants = array(),
  86                                                                 $variantfallbacks = array(), $flags = array(),
  87                                                                 $manualLevel = array() ) {
  88                 global $wgDisabledVariants;
  89                 $this->mLangObj = $langobj;
  90                 $this->mMainLanguageCode = $maincode;
  91                 $this->mVariants = array_diff( $variants, $wgDisabledVariants );
  92                 $this->mVariantFallbacks = $variantfallbacks;
  93                 $this->mVariantNames = Language::fetchLanguageNames();
  94                 $this->mCacheKey = wfMemcKey( 'conversiontables', $maincode );
  95                 $defaultflags = array(
  96                         // 'S' show converted text
  97                         // '+' add rules for alltext
  98                         // 'E' the gave flags is error
  99                         // these flags above are reserved for program
 100                         'A' => 'A',       // add rule for convert code (all text convert)
 101                         'T' => 'T',       // title convert
 102                         'R' => 'R',       // raw content
 103                         'D' => 'D',       // convert description (subclass implement)
 104                         '-' => '-',       // remove convert (not implement)
 105                         'H' => 'H',       // add rule for convert code (but no display in placed code)
 106                         'N' => 'N'        // current variant name
 107                 );
 108                 $this->mFlags = array_merge( $defaultflags, $flags );
 109                 foreach ( $this->mVariants as $v ) {
 110                         if ( array_key_exists( $v, $manualLevel ) ) {
 111                                 $this->mManualLevel[$v] = $manualLevel[$v];
 112                         } else {
 113                                 $this->mManualLevel[$v] = 'bidirectional';
 114                         }
 115                         $this->mFlags[$v] = $v;
 116                 }
 117         }
 118
 119         /**
 120          * Get all valid variants.
 121          * Call this instead of using $this->mVariants directly.
 122          *
 123          * @return Array: contains all valid variants
 124          */
 125         public function getVariants() {
 126                 return $this->mVariants;
 127         }
 128
 129         /**
 130          * In case some variant is not defined in the markup, we need
 131          * to have some fallback. For example, in zh, normally people
 132          * will define zh-hans and zh-hant, but less so for zh-sg or zh-hk.
 133          * when zh-sg is preferred but not defined, we will pick zh-hans
 134          * in this case. Right now this is only used by zh.
 135          *
 136          * @param $variant String: the language code of the variant
 137          * @return String|array: The code of the fallback language or the
 138          *                               main code if there is no fallback
 139          */
 140         public function getVariantFallbacks( $variant ) {
 141                 if ( isset( $this->mVariantFallbacks[$variant] ) ) {
 142                         return $this->mVariantFallbacks[$variant];
 143                 }
 144                 return $this->mMainLanguageCode;
 145         }
 146
 147         /**
 148          * Get the title produced by the conversion rule.
 149          * @return String: The converted title text
 150          */
 151         public function getConvRuleTitle() {
 152                 return $this->mConvRuleTitle;
 153         }
 154
 155         /**
 156          * Get preferred language variant.
 157          * @return String: the preferred language code
 158          */
 159         public function getPreferredVariant() {
 160                 global $wgDefaultLanguageVariant, $wgUser;
 161
 162                 $req = $this->getURLVariant();
 163
 164                 if ( $wgUser->isLoggedIn() && !$req ) {
 165                         $req = $this->getUserVariant();
 166                 } elseif ( !$req ) {
 167                         $req = $this->getHeaderVariant();
 168                 }
 169
 170                 if ( $wgDefaultLanguageVariant && !$req ) {
 171                         $req = $this->validateVariant( $wgDefaultLanguageVariant );
 172                 }
 173
 174                 // This function, unlike the other get*Variant functions, is
 175                 // not memoized (i.e. there return value is not cached) since
 176                 // new information might appear during processing after this
 177                 // is first called.
 178                 if ( $this->validateVariant( $req ) ) {
 179                         return $req;
 180                 }
 181                 return $this->mMainLanguageCode;
 182         }
 183
 184         /**
 185          * Get default variant.
 186          * This function would not be affected by user's settings
 187          * @return String: the default variant code
 188          */
 189         public function getDefaultVariant() {
 190                 global $wgDefaultLanguageVariant;
 191
 192                 $req = $this->getURLVariant();
 193
 194                 if ( !$req ) {
 195                         $req = $this->getHeaderVariant();
 196                 }
 197
 198                 if ( $wgDefaultLanguageVariant && !$req ) {
 199                         $req = $this->validateVariant( $wgDefaultLanguageVariant );
 200                 }
 201
 202                 if ( $req ) {
 203                         return $req;
 204                 }
 205                 return $this->mMainLanguageCode;
 206         }
 207
 208         /**
 209          * Validate the variant
 210          * @param $variant String: the variant to validate
 211          * @return Mixed: returns the variant if it is valid, null otherwise
 212          */
 213         public function validateVariant( $variant = null ) {
 214                 if ( $variant !== null && in_array( $variant, $this->mVariants ) ) {
 215                         return $variant;
 216                 }
 217                 return null;
 218         }
 219
 220         /**
 221          * Get the variant specified in the URL
 222          *
 223          * @return Mixed: variant if one found, false otherwise.
 224          */
 225         public function getURLVariant() {
 226                 global $wgRequest;
 227
 228                 if ( $this->mURLVariant ) {
 229                         return $this->mURLVariant;
 230                 }
 231
 232                 // see if the preference is set in the request
 233                 $ret = $wgRequest->getText( 'variant' );
 234
 235                 if ( !$ret ) {
 236                         $ret = $wgRequest->getVal( 'uselang' );
 237                 }
 238
 239                 return $this->mURLVariant = $this->validateVariant( $ret );
 240         }
 241
 242         /**
 243          * Determine if the user has a variant set.
 244          *
 245          * @return Mixed: variant if one found, false otherwise.
 246          */
 247         protected function getUserVariant() {
 248                 global $wgUser, $wgContLang;
 249
 250                 // memoizing this function wreaks havoc on parserTest.php
 251                 /*
 252                 if ( $this->mUserVariant ) {
 253                         return $this->mUserVariant;
 254                 }
 255                 */
 256
 257                 // Get language variant preference from logged in users
 258                 // Don't call this on stub objects because that causes infinite
 259                 // recursion during initialisation
 260                 if ( $wgUser->isLoggedIn() ) {
 261                         if ( $this->mMainLanguageCode == $wgContLang->getCode() ) {
 262                                 $ret = $wgUser->getOption( 'variant' );
 263                         } else {
 264                                 $ret = $wgUser->getOption( 'variant-' . $this->mMainLanguageCode );
 265                         }
 266                 } else {
 267                         // figure out user lang without constructing wgLang to avoid
 268                         // infinite recursion
 269                         $ret = $wgUser->getOption( 'language' );
 270                 }
 271
 272                 return $this->mUserVariant = $this->validateVariant( $ret );
 273         }
 274
 275         /**
 276          * Determine the language variant from the Accept-Language header.
 277          *
 278          * @return Mixed: variant if one found, false otherwise.
 279          */
 280         protected function getHeaderVariant() {
 281                 global $wgRequest;
 282
 283                 if ( $this->mHeaderVariant ) {
 284                         return $this->mHeaderVariant;
 285                 }
 286
 287                 // see if some supported language variant is set in the
 288                 // HTTP header.
 289                 $languages = array_keys( $wgRequest->getAcceptLang() );
 290                 if ( empty( $languages ) ) {
 291                         return null;
 292                 }
 293
 294                 $fallbackLanguages = array();
 295                 foreach ( $languages as $language ) {
 296                         $this->mHeaderVariant = $this->validateVariant( $language );
 297                         if ( $this->mHeaderVariant ) {
 298                                 break;
 299                         }
 300
 301                         // To see if there are fallbacks of current language.
 302                         // We record these fallback variants, and process
 303                         // them later.
 304                         $fallbacks = $this->getVariantFallbacks( $language );
 305                         if ( is_string( $fallbacks ) && $fallbacks !== $this->mMainLanguageCode ) {
 306                                 $fallbackLanguages[] = $fallbacks;
 307                         } elseif ( is_array( $fallbacks ) ) {
 308                                 $fallbackLanguages =
 309                                         array_merge( $fallbackLanguages, $fallbacks );
 310                         }
 311                 }
 312
 313                 if ( !$this->mHeaderVariant ) {
 314                         // process fallback languages now
 315                         $fallback_languages = array_unique( $fallbackLanguages );
 316                         foreach ( $fallback_languages as $language ) {
 317                                 $this->mHeaderVariant = $this->validateVariant( $language );
 318                                 if ( $this->mHeaderVariant ) {
 319                                         break;
 320                                 }
 321                         }
 322                 }
 323
 324                 return $this->mHeaderVariant;
 325         }
 326
 327         /**
 328          * Dictionary-based conversion.
 329          * This function would not parse the conversion rules.
 330          * If you want to parse rules, try to use convert() or
 331          * convertTo().
 332          *
 333          * @param $text String the text to be converted
 334          * @param $toVariant bool|string the target language code
 335          * @return String the converted text
 336          */
 337         public function autoConvert( $text, $toVariant = false ) {
 338                 wfProfileIn( __METHOD__ );
 339
 340                 $this->loadTables();
 341
 342                 if ( !$toVariant ) {
 343                         $toVariant = $this->getPreferredVariant();
 344                         if ( !$toVariant ) {
 345                                 wfProfileOut( __METHOD__ );
 346                                 return $text;
 347                         }
 348                 }
 349
 350                 if ( $this->guessVariant( $text, $toVariant ) ) {
 351                         wfProfileOut( __METHOD__ );
 352                         return $text;
 353                 }
 354
 355                 /* we convert everything except:
 356                    1. HTML markups (anything between < and >)
 357                    2. HTML entities
 358                    3. placeholders created by the parser
 359                 */
 360                 global $wgParser;
 361                 if ( isset( $wgParser ) && $wgParser->UniqPrefix() != '' ) {
 362                         $marker = '|' . $wgParser->UniqPrefix() . '[\-a-zA-Z0-9]+';
 363                 } else {
 364                         $marker = '';
 365                 }
 366
 367                 // this one is needed when the text is inside an HTML markup
 368                 $htmlfix = '|<[^>]+$|^[^<>]*>';
 369
 370                 // disable convert to variants between <code> tags
 371                 $codefix = '<code>.+?<\/code>|';
 372                 // disable conversion of <script> tags
 373                 $scriptfix = '<script.*?>.*?<\/script>|';
 374                 // disable conversion of <pre> tags
 375                 $prefix = '<pre.*?>.*?<\/pre>|';
 376
 377                 $reg = '/' . $codefix . $scriptfix . $prefix .
 378                         '<[^>]+>|&[a-zA-Z#][a-z0-9]+;' . $marker . $htmlfix . '/s';
 379                 $startPos = 0;
 380                 $sourceBlob = '';
 381                 $literalBlob = '';
 382
 383                 // Guard against delimiter nulls in the input
 384                 $text = str_replace( "\000", '', $text );
 385
 386                 $markupMatches = null;
 387                 $elementMatches = null;
 388                 while ( $startPos < strlen( $text ) ) {
 389                         if ( preg_match( $reg, $text, $markupMatches, PREG_OFFSET_CAPTURE, $startPos ) ) {
 390                                 $elementPos = $markupMatches[0][1];
 391                                 $element = $markupMatches[0][0];
 392                         } else {
 393                                 $elementPos = strlen( $text );
 394                                 $element = '';
 395                         }
 396
 397                         // Queue the part before the markup for translation in a batch
 398                         $sourceBlob .= substr( $text, $startPos, $elementPos - $startPos ) . "\000";
 399
 400                         // Advance to the next position
 401                         $startPos = $elementPos + strlen( $element );
 402
 403                         // Translate any alt or title attributes inside the matched element
 404                         if ( $element !== '' && preg_match( '/^(<[^>\s]*)\s([^>]*)(.*)$/', $element,
 405                                 $elementMatches ) )
 406                         {
 407                                 $attrs = Sanitizer::decodeTagAttributes( $elementMatches[2] );
 408                                 $changed = false;
 409                                 foreach ( array( 'title', 'alt' ) as $attrName ) {
 410                                         if ( !isset( $attrs[$attrName] ) ) {
 411                                                 continue;
 412                                         }
 413                                         $attr = $attrs[$attrName];
 414                                         // Don't convert URLs
 415                                         if ( !strpos( $attr, '://' ) ) {
 416                                                 $attr = $this->recursiveConvertTopLevel( $attr, $toVariant );
 417                                         }
 418
 419                                         // Remove HTML tags to avoid disrupting the layout
 420                                         $attr = preg_replace( '/<[^>]+>/', '', $attr );
 421                                         if ( $attr !== $attrs[$attrName] ) {
 422                                                 $attrs[$attrName] = $attr;
 423                                                 $changed = true;
 424                                         }
 425                                 }
 426                                 if ( $changed ) {
 427                                         $element = $elementMatches[1] . Html::expandAttributes( $attrs ) .
 428                                                 $elementMatches[3];
 429                                 }
 430                         }
 431                         $literalBlob .= $element . "\000";
 432                 }
 433
 434                 // Do the main translation batch
 435                 $translatedBlob = $this->translate( $sourceBlob, $toVariant );
 436
 437                 // Put the output back together
 438                 $translatedIter = StringUtils::explode( "\000", $translatedBlob );
 439                 $literalIter = StringUtils::explode( "\000", $literalBlob );
 440                 $output = '';
 441                 while ( $translatedIter->valid() && $literalIter->valid() ) {
 442                         $output .= $translatedIter->current();
 443                         $output .= $literalIter->current();
 444                         $translatedIter->next();
 445                         $literalIter->next();
 446                 }
 447
 448                 wfProfileOut( __METHOD__ );
 449                 return $output;
 450         }
 451
 452         /**
 453          * Translate a string to a variant.
 454          * Doesn't parse rules or do any of that other stuff, for that use
 455          * convert() or convertTo().
 456          *
 457          * @param $text String: text to convert
 458          * @param $variant String: variant language code
 459          * @return String: translated text
 460          */
 461         public function translate( $text, $variant ) {
 462                 wfProfileIn( __METHOD__ );
 463                 // If $text is empty or only includes spaces, do nothing
 464                 // Otherwise translate it
 465                 if ( trim( $text ) ) {
 466                         $this->loadTables();
 467                         $text = $this->mTables[$variant]->replace( $text );
 468                 }
 469                 wfProfileOut( __METHOD__ );
 470                 return $text;
 471         }
 472
 473         /**
 474          * Call translate() to convert text to all valid variants.
 475          *
 476          * @param $text String: the text to be converted
 477          * @return Array: variant => converted text
 478          */
 479         public function autoConvertToAllVariants( $text ) {
 480                 wfProfileIn( __METHOD__ );
 481                 $this->loadTables();
 482
 483                 $ret = array();
 484                 foreach ( $this->mVariants as $variant ) {
 485                         $ret[$variant] = $this->translate( $text, $variant );
 486                 }
 487
 488                 wfProfileOut( __METHOD__ );
 489                 return $ret;
 490         }
 491
 492         /**
 493          * Convert link text to all valid variants.
 494          * In the first, this function only convert text outside the
 495          * "-{" "}-" markups. Since the "{" and "}" are not allowed in
 496          * titles, the text will get all converted always.
 497          * So I removed this feature and deprecated the function.
 498          *
 499          * @param $text String: the text to be converted
 500          * @return Array: variant => converted text
 501          * @deprecated since 1.17 Use autoConvertToAllVariants() instead
 502          */
 503         public function convertLinkToAllVariants( $text ) {
 504                 return $this->autoConvertToAllVariants( $text );
 505         }
 506
 507         /**
 508          * Apply manual conversion rules.
 509          *
 510          * @param $convRule ConverterRule Object of ConverterRule
 511          */
 512         protected function applyManualConv( $convRule ) {
 513                 // Use syntax -{T|zh-cn:TitleCN; zh-tw:TitleTw}- to custom
 514                 // title conversion.
 515                 // Bug 24072: $mConvRuleTitle was overwritten by other manual
 516                 // rule(s) not for title, this breaks the title conversion.
 517                 $newConvRuleTitle = $convRule->getTitle();
 518                 if ( $newConvRuleTitle ) {
 519                         // So I add an empty check for getTitle()
 520                         $this->mConvRuleTitle = $newConvRuleTitle;
 521                 }
 522
 523                 // merge/remove manual conversion rules to/from global table
 524                 $convTable = $convRule->getConvTable();
 525                 $action = $convRule->getRulesAction();
 526                 foreach ( $convTable as $variant => $pair ) {
 527                         if ( !$this->validateVariant( $variant ) ) {
 528                                 continue;
 529                         }
 530
 531                         if ( $action == 'add' ) {
 532                                 foreach ( $pair as $from => $to ) {
 533                                         // to ensure that $from and $to not be left blank
 534                                         // so $this->translate() could always return a string
 535                                         if ( $from || $to ) {
 536                                                 // more efficient than array_merge(), about 2.5 times.
 537                                                 $this->mTables[$variant]->setPair( $from, $to );
 538                                         }
 539                                 }
 540                         } elseif ( $action == 'remove' ) {
 541                                 $this->mTables[$variant]->removeArray( $pair );
 542                         }
 543                 }
 544         }
 545
 546         /**
 547          * Auto convert a Title object to a readable string in the
 548          * preferred variant.
 549          *
 550          * @param $title Title a object of Title
 551          * @return String: converted title text
 552          */
 553         public function convertTitle( $title ) {
 554                 $variant = $this->getPreferredVariant();
 555                 $index = $title->getNamespace();
 556                 if ( $index !== NS_MAIN ) {
 557                         $text = $this->convertNamespace( $index, $variant ) . ':';
 558                 } else {
 559                         $text = '';
 560                 }
 561                 $text .= $this->translate( $title->getText(), $variant );
 562                 return $text;
 563         }
 564
 565         /**
 566          * Get the namespace display name in the preferred variant.
 567          *
 568          * @param $index int namespace id
 569          * @param $variant string|null variant code or null for preferred variant
 570          * @return String: namespace name for display
 571          */
 572         public function convertNamespace( $index, $variant = null ) {
 573                 if ( $variant === null ) {
 574                         $variant = $this->getPreferredVariant();
 575                 }
 576                 if ( $index === NS_MAIN ) {
 577                         return '';
 578                 } else {
 579                         // First check if a message gives a converted name in the target variant.
 580                         $nsConvMsg = wfMessage( 'conversion-ns' . $index )->inLanguage( $variant );
 581                         if ( $nsConvMsg->exists() ) {
 582                                 return $nsConvMsg->plain();
 583                         }
 584                         // Then check if a message gives a converted name in content language
 585                         // which needs extra translation to the target variant.
 586                         $nsConvMsg = wfMessage( 'conversion-ns' . $index )->inContentLanguage();
 587                         if ( $nsConvMsg->exists() ) {
 588                                 return $this->translate( $nsConvMsg->plain(), $variant );
 589                         }
 590                         // No message exists, retrieve it from the target variant's namespace names.
 591                         $langObj = $this->mLangObj->factory( $variant );
 592                         return $langObj->getFormattedNsText( $index );
 593                 }
 594         }
 595
 596         /**
 597          * Convert text to different variants of a language. The automatic
 598          * conversion is done in autoConvert(). Here we parse the text
 599          * marked with -{}-, which specifies special conversions of the
 600          * text that can not be accomplished in autoConvert().
 601          *
 602          * Syntax of the markup:
 603          * -{code1:text1;code2:text2;...}-  or
 604          * -{flags|code1:text1;code2:text2;...}-  or
 605          * -{text}- in which case no conversion should take place for text
 606          *
 607          * @param $text String: text to be converted
 608          * @return String: converted text
 609          */
 610         public function convert( $text ) {
 611                 $variant = $this->getPreferredVariant();
 612                 return $this->convertTo( $text, $variant );
 613         }
 614
 615         /**
 616          * Same as convert() except a extra parameter to custom variant.
 617          *
 618          * @param $text String: text to be converted
 619          * @param $variant String: the target variant code
 620          * @return String: converted text
 621          */
 622         public function convertTo( $text, $variant ) {
 623                 global $wgDisableLangConversion;
 624                 if ( $wgDisableLangConversion ) {
 625                         return $text;
 626                 }
 627                 // Reset converter state for a new converter run.
 628                 $this->mConvRuleTitle = false;
 629                 return $this->recursiveConvertTopLevel( $text, $variant );
 630         }
 631
 632         /**
 633          * Recursively convert text on the outside. Allow to use nested
 634          * markups to custom rules.
 635          *
 636          * @param $text String: text to be converted
 637          * @param $variant String: the target variant code
 638          * @param $depth Integer: depth of recursion
 639          * @return String: converted text
 640          */
 641         protected function recursiveConvertTopLevel( $text, $variant, $depth = 0 ) {
 642                 $startPos = 0;
 643                 $out = '';
 644                 $length = strlen( $text );
 645                 $shouldConvert = !$this->guessVariant( $text, $variant );
 646
 647                 while ( $startPos < $length ) {
 648                         $pos = strpos( $text, '-{', $startPos );
 649
 650                         if ( $pos === false ) {
 651                                 // No more markup, append final segment
 652                                 $fragment = substr( $text, $startPos );
 653                                 $out .= $shouldConvert ? $this->autoConvert( $fragment, $variant ) : $fragment;
 654                                 return $out;
 655                         }
 656
 657                         // Markup found
 658                         // Append initial segment
 659                         $fragment = substr( $text, $startPos, $pos - $startPos );
 660                         $out .= $shouldConvert ? $this->autoConvert( $fragment, $variant ) : $fragment;
 661
 662                         // Advance position
 663                         $startPos = $pos;
 664
 665                         // Do recursive conversion
 666                         $out .= $this->recursiveConvertRule( $text, $variant, $startPos, $depth + 1 );
 667                 }
 668
 669                 return $out;
 670         }
 671
 672         /**
 673          * Recursively convert text on the inside.
 674          *
 675          * @param $text String: text to be converted
 676          * @param $variant String: the target variant code
 677          * @param $startPos int
 678          * @param $depth Integer: depth of recursion
 679          *
 680          * @throws MWException
 681          * @return String: converted text
 682          */
 683         protected function recursiveConvertRule( $text, $variant, &$startPos, $depth = 0 ) {
 684                 // Quick sanity check (no function calls)
 685                 if ( $text[$startPos] !== '-' || $text[$startPos + 1] !== '{' ) {
 686                         throw new MWException( __METHOD__ . ': invalid input string' );
 687                 }
 688
 689                 $startPos += 2;
 690                 $inner = '';
 691                 $warningDone = false;
 692                 $length = strlen( $text );
 693
 694                 while ( $startPos < $length ) {
 695                         $m = false;
 696                         preg_match( '/-\{|\}-/', $text, $m, PREG_OFFSET_CAPTURE, $startPos );
 697                         if ( !$m ) {
 698                                 // Unclosed rule
 699                                 break;
 700                         }
 701
 702                         $token = $m[0][0];
 703                         $pos = $m[0][1];
 704
 705                         // Markup found
 706                         // Append initial segment
 707                         $inner .= substr( $text, $startPos, $pos - $startPos );
 708
 709                         // Advance position
 710                         $startPos = $pos;
 711
 712                         switch ( $token ) {
 713                                 case '-{':
 714                                         // Check max depth
 715                                         if ( $depth >= $this->mMaxDepth ) {
 716                                                 $inner .= '-{';
 717                                                 if ( !$warningDone ) {
 718                                                         $inner .= '<span class="error">' .
 719                                                                 wfMessage( 'language-converter-depth-warning' )
 720                                                                         ->numParams( $this->mMaxDepth )->inContentLanguage()->text() .
 721                                                                 '</span>';
 722                                                         $warningDone = true;
 723                                                 }
 724                                                 $startPos += 2;
 725                                                 continue;
 726                                         }
 727                                         // Recursively parse another rule
 728                                         $inner .= $this->recursiveConvertRule( $text, $variant, $startPos, $depth + 1 );
 729                                         break;
 730                                 case '}-':
 731                                         // Apply the rule
 732                                         $startPos += 2;
 733                                         $rule = new ConverterRule( $inner, $this );
 734                                         $rule->parse( $variant );
 735                                         $this->applyManualConv( $rule );
 736                                         return $rule->getDisplay();
 737                                 default:
 738                                         throw new MWException( __METHOD__ . ': invalid regex match' );
 739                         }
 740                 }
 741
 742                 // Unclosed rule
 743                 if ( $startPos < $length ) {
 744                         $inner .= substr( $text, $startPos );
 745                 }
 746                 $startPos = $length;
 747                 return '-{' . $this->autoConvert( $inner, $variant );
 748         }
 749
 750         /**
 751          * If a language supports multiple variants, it is possible that
 752          * non-existing link in one variant actually exists in another variant.
 753          * This function tries to find it. See e.g. LanguageZh.php
 754          *
 755          * @param $link String: the name of the link
 756          * @param $nt Mixed: the title object of the link
 757          * @param $ignoreOtherCond Boolean: to disable other conditions when
 758          *              we need to transclude a template or update a category's link
 759          * @return Null, the input parameters may be modified upon return
 760          */
 761         public function findVariantLink( &$link, &$nt, $ignoreOtherCond = false ) {
 762                 # If the article has already existed, there is no need to
 763                 # check it again, otherwise it may cause a fault.
 764                 if ( is_object( $nt ) && $nt->exists() ) {
 765                         return;
 766                 }
 767
 768                 global $wgDisableLangConversion, $wgDisableTitleConversion, $wgRequest,
 769                         $wgUser;
 770                 $isredir = $wgRequest->getText( 'redirect', 'yes' );
 771                 $action = $wgRequest->getText( 'action' );
 772                 $linkconvert = $wgRequest->getText( 'linkconvert', 'yes' );
 773                 $disableLinkConversion = $wgDisableLangConversion
 774                         || $wgDisableTitleConversion;
 775                 $linkBatch = new LinkBatch();
 776
 777                 $ns = NS_MAIN;
 778
 779                 if ( $disableLinkConversion ||
 780                         ( !$ignoreOtherCond &&
 781                                 ( $isredir == 'no'
 782                                         || $action == 'edit'
 783                                         || $action == 'submit'
 784                                         || $linkconvert == 'no'
 785                                         || $wgUser->getOption( 'noconvertlink' ) == 1 ) ) ) {
 786                         return;
 787                 }
 788
 789                 if ( is_object( $nt ) ) {
 790                         $ns = $nt->getNamespace();
 791                 }
 792
 793                 $variants = $this->autoConvertToAllVariants( $link );
 794                 if ( !$variants ) { // give up
 795                         return;
 796                 }
 797
 798                 $titles = array();
 799
 800                 foreach ( $variants as $v ) {
 801                         if ( $v != $link ) {
 802                                 $varnt = Title::newFromText( $v, $ns );
 803                                 if ( !is_null( $varnt ) ) {
 804                                         $linkBatch->addObj( $varnt );
 805                                         $titles[] = $varnt;
 806                                 }
 807                         }
 808                 }
 809
 810                 // fetch all variants in single query
 811                 $linkBatch->execute();
 812
 813                 foreach ( $titles as $varnt ) {
 814                         if ( $varnt->getArticleID() > 0 ) {
 815                                 $nt = $varnt;
 816                                 $link = $varnt->getText();
 817                                 break;
 818                         }
 819                 }
 820         }
 821
 822         /**
 823          * Returns language specific hash options.
 824          *
 825          * @return string
 826          */
 827         public function getExtraHashOptions() {
 828                 $variant = $this->getPreferredVariant();
 829                 return '!' . $variant;
 830         }
 831
 832         /**
 833          * Guess if a text is written in a variant. This should be implemented in subclasses.
 834          *
 835          * @param string $text the text to be checked
 836          * @param string $variant language code of the variant to be checked for
 837          * @return bool true if $text appears to be written in $variant, false if not
 838          *
 839          * @author Nikola Smolenski <smolensk@eunet.rs>
 840          * @since 1.19
 841          */
 842         public function guessVariant( $text, $variant ) {
 843                 return false;
 844         }
 845
 846         /**
 847          * Load default conversion tables.
 848          * This method must be implemented in derived class.
 849          *
 850          * @private
 851          * @throws MWException
 852          */
 853         function loadDefaultTables() {
 854                 $name = get_class( $this );
 855                 throw new MWException( "Must implement loadDefaultTables() method in class $name" );
 856         }
 857
 858         /**
 859          * Load conversion tables either from the cache or the disk.
 860          * @private
 861          * @param $fromCache Boolean: load from memcached? Defaults to true.
 862          */
 863         function loadTables( $fromCache = true ) {
 864                 global $wgLangConvMemc;
 865
 866                 if ( $this->mTablesLoaded ) {
 867                         return;
 868                 }
 869
 870                 wfProfileIn( __METHOD__ );
 871                 $this->mTablesLoaded = true;
 872                 $this->mTables = false;
 873                 if ( $fromCache ) {
 874                         wfProfileIn( __METHOD__ . '-cache' );
 875                         $this->mTables = $wgLangConvMemc->get( $this->mCacheKey );
 876                         wfProfileOut( __METHOD__ . '-cache' );
 877                 }
 878                 if ( !$this->mTables || !array_key_exists( self::CACHE_VERSION_KEY, $this->mTables ) ) {
 879                         wfProfileIn( __METHOD__ . '-recache' );
 880                         // not in cache, or we need a fresh reload.
 881                         // We will first load the default tables
 882                         // then update them using things in MediaWiki:Conversiontable/*
 883                         $this->loadDefaultTables();
 884                         foreach ( $this->mVariants as $var ) {
 885                                 $cached = $this->parseCachedTable( $var );
 886                                 $this->mTables[$var]->mergeArray( $cached );
 887                         }
 888
 889                         $this->postLoadTables();
 890                         $this->mTables[self::CACHE_VERSION_KEY] = true;
 891
 892                         $wgLangConvMemc->set( $this->mCacheKey, $this->mTables, 43200 );
 893                         wfProfileOut( __METHOD__ . '-recache' );
 894                 }
 895                 wfProfileOut( __METHOD__ );
 896         }
 897
 898         /**
 899          * Hook for post processing after conversion tables are loaded.
 900          */
 901         function postLoadTables() { }
 902
 903         /**
 904          * Reload the conversion tables.
 905          *
 906          * @private
 907          */
 908         function reloadTables() {
 909                 if ( $this->mTables ) {
 910                         unset( $this->mTables );
 911                 }
 912                 $this->mTablesLoaded = false;
 913                 $this->loadTables( false );
 914         }
 915
 916         /**
 917          * Parse the conversion table stored in the cache.
 918          *
 919          * The tables should be in blocks of the following form:
 920          *              -{
 921          *                      word => word ;
 922          *                      word => word ;
 923          *                      ...
 924          *              }-
 925          *
 926          * To make the tables more manageable, subpages are allowed
 927          * and will be parsed recursively if $recursive == true.
 928          *
 929          * @param $code String: language code
 930          * @param $subpage String: subpage name
 931          * @param $recursive Boolean: parse subpages recursively? Defaults to true.
 932          *
 933          * @return array
 934          */
 935         function parseCachedTable( $code, $subpage = '', $recursive = true ) {
 936                 static $parsed = array();
 937
 938                 $key = 'Conversiontable/' . $code;
 939                 if ( $subpage ) {
 940                         $key .= '/' . $subpage;
 941                 }
 942                 if ( array_key_exists( $key, $parsed ) ) {
 943                         return array();
 944                 }
 945
 946                 $parsed[$key] = true;
 947
 948                 if ( $subpage === '' ) {
 949                         $txt = MessageCache::singleton()->getMsgFromNamespace( $key, $code );
 950                 } else {
 951                         $txt = false;
 952                         $title = Title::makeTitleSafe( NS_MEDIAWIKI, $key );
 953                         if ( $title && $title->exists() ) {
 954                                 $revision = Revision::newFromTitle( $title );
 955                                 if ( $revision ) {
 956                                         if ( $revision->getContentModel() == CONTENT_MODEL_WIKITEXT ) {
 957                                                 $txt = $revision->getContent( Revision::RAW )->getNativeData();
 958                                         }
 959
 960                                         // @todo in the future, use a specialized content model, perhaps based on json!
 961                                 }
 962                         }
 963                 }
 964
 965                 # Nothing to parse if there's no text
 966                 if ( $txt === false || $txt === null || $txt === '' ) {
 967                         return array();
 968                 }
 969
 970                 // get all subpage links of the form
 971                 // [[MediaWiki:Conversiontable/zh-xx/...|...]]
 972                 $linkhead = $this->mLangObj->getNsText( NS_MEDIAWIKI ) .
 973                         ':Conversiontable';
 974                 $subs = StringUtils::explode( '[[', $txt );
 975                 $sublinks = array();
 976                 foreach ( $subs as $sub ) {
 977                         $link = explode( ']]', $sub, 2 );
 978                         if ( count( $link ) != 2 ) {
 979                                 continue;
 980                         }
 981                         $b = explode( '|', $link[0], 2 );
 982                         $b = explode( '/', trim( $b[0] ), 3 );
 983                         if ( count( $b ) == 3 ) {
 984                                 $sublink = $b[2];
 985                         } else {
 986                                 $sublink = '';
 987                         }
 988
 989                         if ( $b[0] == $linkhead && $b[1] == $code ) {
 990                                 $sublinks[] = $sublink;
 991                         }
 992                 }
 993
 994                 // parse the mappings in this page
 995                 $blocks = StringUtils::explode( '-{', $txt );
 996                 $ret = array();
 997                 $first = true;
 998                 foreach ( $blocks as $block ) {
 999                         if ( $first ) {
1000                                 // Skip the part before the first -{
1001                                 $first = false;
1002                                 continue;
1003                         }
1004                         $mappings = explode( '}-', $block, 2 );
1005                         $stripped = str_replace( array( "'", '"', '*', '#' ), '', $mappings[0] );
1006                         $table = StringUtils::explode( ';', $stripped );
1007                         foreach ( $table as $t ) {
1008                                 $m = explode( '=>', $t, 3 );
1009                                 if ( count( $m ) != 2 ) {
1010                                         continue;
1011                                 }
1012                                 // trim any trailling comments starting with '//'
1013                                 $tt = explode( '//', $m[1], 2 );
1014                                 $ret[trim( $m[0] )] = trim( $tt[0] );
1015                         }
1016                 }
1017
1018                 // recursively parse the subpages
1019                 if ( $recursive ) {
1020                         foreach ( $sublinks as $link ) {
1021                                 $s = $this->parseCachedTable( $code, $link, $recursive );
1022                                 $ret = array_merge( $ret, $s );
1023                         }
1024                 }
1025
1026                 if ( $this->mUcfirst ) {
1027                         foreach ( $ret as $k => $v ) {
1028                                 $ret[$this->mLangObj->ucfirst( $k )] = $this->mLangObj->ucfirst( $v );
1029                         }
1030                 }
1031                 return $ret;
1032         }
1033
1034         /**
1035          * Enclose a string with the "no conversion" tag. This is used by
1036          * various functions in the Parser.
1037          *
1038          * @param $text String: text to be tagged for no conversion
1039          * @param $noParse Boolean: unused
1040          * @return String: the tagged text
1041          */
1042         public function markNoConversion( $text, $noParse = false ) {
1043                 # don't mark if already marked
1044                 if ( strpos( $text, '-{' ) || strpos( $text, '}-' ) ) {
1045                         return $text;
1046                 }
1047
1048                 $ret = "-{R|$text}-";
1049                 return $ret;
1050         }
1051
1052         /**
1053          * Convert the sorting key for category links. This should make different
1054          * keys that are variants of each other map to the same key.
1055          *
1056          * @param $key string
1057          *
1058          * @return string
1059          */
1060         function convertCategoryKey( $key ) {
1061                 return $key;
1062         }
1063
1064         /**
1065          * Hook to refresh the cache of conversion tables when
1066          * MediaWiki:Conversiontable* is updated.
1067          * @private
1068          *
1069          * @param $page WikiPage object
1070          * @param $user Object: User object for the current user
1071          * @param $content Content: new page content
1072          * @param $summary String: edit summary of the edit
1073          * @param $isMinor Boolean: was the edit marked as minor?
1074          * @param $isWatch Boolean: did the user watch this page or not?
1075          * @param $section
1076          * @param $flags int Bitfield
1077          * @param $revision Object: new Revision object or null
1078          * @return Boolean: true
1079          */
1080         function OnPageContentSaveComplete( $page, $user, $content, $summary, $isMinor,
1081                         $isWatch, $section, $flags, $revision ) {
1082                 $titleobj = $page->getTitle();
1083                 if ( $titleobj->getNamespace() == NS_MEDIAWIKI ) {
1084                         $title = $titleobj->getDBkey();
1085                         $t = explode( '/', $title, 3 );
1086                         $c = count( $t );
1087                         if ( $c > 1 && $t[0] == 'Conversiontable' ) {
1088                                 if ( $this->validateVariant( $t[1] ) ) {
1089                                         $this->reloadTables();
1090                                 }
1091                         }
1092                 }
1093                 return true;
1094         }
1095
1096         /**
1097          * Armour rendered math against conversion.
1098          * Escape special chars in parsed math text. (in most cases are img elements)
1099          *
1100          * @param $text String: text to armour against conversion
1101          * @return String: armoured text where { and } have been converted to
1102          *                 &#123; and &#125;
1103          * @deprecated since 1.22 is no longer used
1104          */
1105         public function armourMath( $text ) {
1106                 // convert '-{' and '}-' to '-&#123;' and '&#125;-' to prevent
1107                 // any unwanted markup appearing in the math image tag.
1108                 $text = strtr( $text, array( '-{' => '-&#123;', '}-' => '&#125;-' ) );
1109                 return $text;
1110         }
1111
1112         /**
1113          * Get the cached separator pattern for ConverterRule::parseRules()
1114          */
1115         function getVarSeparatorPattern() {
1116                 if ( is_null( $this->mVarSeparatorPattern ) ) {
1117                         // varsep_pattern for preg_split:
1118                         // text should be splited by ";" only if a valid variant
1119                         // name exist after the markup, for example:
1120                         //  -{zh-hans:<span style="font-size:120%;">xxx</span>;zh-hant:\
1121                         //      <span style="font-size:120%;">yyy</span>;}-
1122                         // we should split it as:
1123                         //  array(
1124                         //        [0] => 'zh-hans:<span style="font-size:120%;">xxx</span>'
1125                         //        [1] => 'zh-hant:<span style="font-size:120%;">yyy</span>'
1126                         //        [2] => ''
1127                         //       )
1128                         $pat = '/;\s*(?=';
1129                         foreach ( $this->mVariants as $variant ) {
1130                                 // zh-hans:xxx;zh-hant:yyy
1131                                 $pat .= $variant . '\s*:|';
1132                                 // xxx=>zh-hans:yyy; xxx=>zh-hant:zzz
1133                                 $pat .= '[^;]*?=>\s*' . $variant . '\s*:|';
1134                         }
1135                         $pat .= '\s*$)/';
1136                         $this->mVarSeparatorPattern = $pat;
1137                 }
1138                 return $this->mVarSeparatorPattern;
1139         }
1140 }
1141
1142 /**
1143  * Parser for rules of language conversion , parse rules in -{ }- tag.
1144  * @ingroup Language
1145  * @author fdcn <fdcn64@gmail.com>, PhiLiP <philip.npc@gmail.com>
1146  */
1147 class ConverterRule {
1148         public $mText; // original text in -{text}-
1149         public $mConverter; // LanguageConverter object
1150         public $mRuleDisplay = '';
1151         public $mRuleTitle = false;
1152         public $mRules = '';// string : the text of the rules
1153         public $mRulesAction = 'none';
1154         public $mFlags = array();
1155         public $mVariantFlags = array();
1156         public $mConvTable = array();
1157         public $mBidtable = array();// array of the translation in each variant
1158         public $mUnidtable = array();// array of the translation in each variant
1159
1160         /**
1161          * Constructor
1162          *
1163          * @param $text String: the text between -{ and }-
1164          * @param $converter LanguageConverter object
1165          */
1166         public function __construct( $text, $converter ) {
1167                 $this->mText = $text;
1168                 $this->mConverter = $converter;
1169         }
1170
1171         /**
1172          * Check if variants array in convert array.
1173          *
1174          * @param $variants Array or string: variant language code
1175          * @return String: translated text
1176          */
1177         public function getTextInBidtable( $variants ) {
1178                 $variants = (array)$variants;
1179                 if ( !$variants ) {
1180                         return false;
1181                 }
1182                 foreach ( $variants as $variant ) {
1183                         if ( isset( $this->mBidtable[$variant] ) ) {
1184                                 return $this->mBidtable[$variant];
1185                         }
1186                 }
1187                 return false;
1188         }
1189
1190         /**
1191          * Parse flags with syntax -{FLAG| ... }-
1192          * @private
1193          */
1194         function parseFlags() {
1195                 $text = $this->mText;
1196                 $flags = array();
1197                 $variantFlags = array();
1198
1199                 $sepPos = strpos( $text, '|' );
1200                 if ( $sepPos !== false ) {
1201                         $validFlags = $this->mConverter->mFlags;
1202                         $f = StringUtils::explode( ';', substr( $text, 0, $sepPos ) );
1203                         foreach ( $f as $ff ) {
1204                                 $ff = trim( $ff );
1205                                 if ( isset( $validFlags[$ff] ) ) {
1206                                         $flags[$validFlags[$ff]] = true;
1207                                 }
1208                         }
1209                         $text = strval( substr( $text, $sepPos + 1 ) );
1210                 }
1211
1212                 if ( !$flags ) {
1213                         $flags['S'] = true;
1214                 } elseif ( isset( $flags['R'] ) ) {
1215                         $flags = array( 'R' => true );// remove other flags
1216                 } elseif ( isset( $flags['N'] ) ) {
1217                         $flags = array( 'N' => true );// remove other flags
1218                 } elseif ( isset( $flags['-'] ) ) {
1219                         $flags = array( '-' => true );// remove other flags
1220                 } elseif ( count( $flags ) == 1 && isset( $flags['T'] ) ) {
1221                         $flags['H'] = true;
1222                 } elseif ( isset( $flags['H'] ) ) {
1223                         // replace A flag, and remove other flags except T
1224                         $temp = array( '+' => true, 'H' => true );
1225                         if ( isset( $flags['T'] ) ) {
1226                                 $temp['T'] = true;
1227                         }
1228                         if ( isset( $flags['D'] ) ) {
1229                                 $temp['D'] = true;
1230                         }
1231                         $flags = $temp;
1232                 } else {
1233                         if ( isset( $flags['A'] ) ) {
1234                                 $flags['+'] = true;
1235                                 $flags['S'] = true;
1236                         }
1237                         if ( isset( $flags['D'] ) ) {
1238                                 unset( $flags['S'] );
1239                         }
1240                         // try to find flags like "zh-hans", "zh-hant"
1241                         // allow syntaxes like "-{zh-hans;zh-hant|XXXX}-"
1242                         $variantFlags = array_intersect( array_keys( $flags ), $this->mConverter->mVariants );
1243                         if ( $variantFlags ) {
1244                                 $variantFlags = array_flip( $variantFlags );
1245                                 $flags = array();
1246                         }
1247                 }
1248                 $this->mVariantFlags = $variantFlags;
1249                 $this->mRules = $text;
1250                 $this->mFlags = $flags;
1251         }
1252
1253         /**
1254          * Generate conversion table.
1255          * @private
1256          */
1257         function parseRules() {
1258                 $rules = $this->mRules;
1259                 $bidtable = array();
1260                 $unidtable = array();
1261                 $variants = $this->mConverter->mVariants;
1262                 $varsep_pattern = $this->mConverter->getVarSeparatorPattern();
1263
1264                 // Split according to $varsep_pattern, but ignore semicolons from HTML entities
1265                 $rules = preg_replace( '/(&[#a-zA-Z0-9]+);/', "$1\x01", $rules );
1266                 $choice = preg_split( $varsep_pattern, $rules );
1267                 $choice = str_replace( "\x01", ';', $choice );
1268
1269                 foreach ( $choice as $c ) {
1270                         $v = explode( ':', $c, 2 );
1271                         if ( count( $v ) != 2 ) {
1272                                 // syntax error, skip
1273                                 continue;
1274                         }
1275                         $to = trim( $v[1] );
1276                         $v = trim( $v[0] );
1277                         $u = explode( '=>', $v, 2 );
1278                         // if $to is empty, strtr() could return a wrong result
1279                         if ( count( $u ) == 1 && $to && in_array( $v, $variants ) ) {
1280                                 $bidtable[$v] = $to;
1281                         } elseif ( count( $u ) == 2 ) {
1282                                 $from = trim( $u[0] );
1283                                 $v = trim( $u[1] );
1284                                 if ( array_key_exists( $v, $unidtable )
1285                                         && !is_array( $unidtable[$v] )
1286                                         && $to
1287                                         && in_array( $v, $variants ) ) {
1288                                         $unidtable[$v] = array( $from => $to );
1289                                 } elseif ( $to && in_array( $v, $variants ) ) {
1290                                         $unidtable[$v][$from] = $to;
1291                                 }
1292                         }
1293                         // syntax error, pass
1294                         if ( !isset( $this->mConverter->mVariantNames[$v] ) ) {
1295                                 $bidtable = array();
1296                                 $unidtable = array();
1297                                 break;
1298                         }
1299                 }
1300                 $this->mBidtable = $bidtable;
1301                 $this->mUnidtable = $unidtable;
1302         }
1303
1304         /**
1305          * @private
1306          *
1307          * @return string
1308          */
1309         function getRulesDesc() {
1310                 $codesep = $this->mConverter->mDescCodeSep;
1311                 $varsep = $this->mConverter->mDescVarSep;
1312                 $text = '';
1313                 foreach ( $this->mBidtable as $k => $v ) {
1314                         $text .= $this->mConverter->mVariantNames[$k] . "$codesep$v$varsep";
1315                 }
1316                 foreach ( $this->mUnidtable as $k => $a ) {
1317                         foreach ( $a as $from => $to ) {
1318                                 $text .= $from . '⇒' . $this->mConverter->mVariantNames[$k] .
1319                                         "$codesep$to$varsep";
1320                         }
1321                 }
1322                 return $text;
1323         }
1324
1325         /**
1326          * Parse rules conversion.
1327          * @private
1328          *
1329          * @param $variant
1330          *
1331          * @return string
1332          */
1333         function getRuleConvertedStr( $variant ) {
1334                 $bidtable = $this->mBidtable;
1335                 $unidtable = $this->mUnidtable;
1336
1337                 if ( count( $bidtable ) + count( $unidtable ) == 0 ) {
1338                         return $this->mRules;
1339                 } else {
1340                         // display current variant in bidirectional array
1341                         $disp = $this->getTextInBidtable( $variant );
1342                         // or display current variant in fallbacks
1343                         if ( !$disp ) {
1344                                 $disp = $this->getTextInBidtable(
1345                                                 $this->mConverter->getVariantFallbacks( $variant ) );
1346                         }
1347                         // or display current variant in unidirectional array
1348                         if ( !$disp && array_key_exists( $variant, $unidtable ) ) {
1349                                 $disp = array_values( $unidtable[$variant] );
1350                                 $disp = $disp[0];
1351                         }
1352                         // or display frist text under disable manual convert
1353                         if ( !$disp && $this->mConverter->mManualLevel[$variant] == 'disable' ) {
1354                                 if ( count( $bidtable ) > 0 ) {
1355                                         $disp = array_values( $bidtable );
1356                                         $disp = $disp[0];
1357                                 } else {
1358                                         $disp = array_values( $unidtable );
1359                                         $disp = array_values( $disp[0] );
1360                                         $disp = $disp[0];
1361                                 }
1362                         }
1363                         return $disp;
1364                 }
1365         }
1366
1367         /**
1368          * Similar to getRuleConvertedStr(), but this prefers to use original
1369          * page title if $variant === $this->mConverter->mMainLanguageCode
1370          * and may return false in this case (so this title conversion rule
1371          * will be ignored and the original title is shown).
1372          *
1373          * @since 1.22
1374          * @param $variant The variant code to display page title in
1375          * @return String|false The converted title or false if just page name
1376          */
1377         function getRuleConvertedTitle( $variant ) {
1378                 if ( $variant === $this->mConverter->mMainLanguageCode ) {
1379                         // If a string targeting exactly this variant is set,
1380                         // use it. Otherwise, just return false, so the real
1381                         // page name can be shown (and because variant === main,
1382                         // there'll be no further automatic conversion).
1383                         $disp = $this->getTextInBidtable( $variant );
1384                         if ( $disp ) {
1385                                 return $disp;
1386                         }
1387                         if ( array_key_exists( $variant, $this->mUnidtable ) ) {
1388                                 $disp = array_values( $this->mUnidtable[$variant] );
1389                                 $disp = $disp[0];
1390                         }
1391                         // Assigned above or still false.
1392                         return $disp;
1393                 } else {
1394                         return $this->getRuleConvertedStr( $variant );
1395                 }
1396         }
1397
1398         /**
1399          * Generate conversion table for all text.
1400          * @private
1401          */
1402         function generateConvTable() {
1403                 // Special case optimisation
1404                 if ( !$this->mBidtable && !$this->mUnidtable ) {
1405                         $this->mConvTable = array();
1406                         return;
1407                 }
1408
1409                 $bidtable = $this->mBidtable;
1410                 $unidtable = $this->mUnidtable;
1411                 $manLevel = $this->mConverter->mManualLevel;
1412
1413                 $vmarked = array();
1414                 foreach ( $this->mConverter->mVariants as $v ) {
1415                         /* for bidirectional array
1416                                 fill in the missing variants, if any,
1417                                 with fallbacks */
1418                         if ( !isset( $bidtable[$v] ) ) {
1419                                 $variantFallbacks =
1420                                         $this->mConverter->getVariantFallbacks( $v );
1421                                 $vf = $this->getTextInBidtable( $variantFallbacks );
1422                                 if ( $vf ) {
1423                                         $bidtable[$v] = $vf;
1424                                 }
1425                         }
1426
1427                         if ( isset( $bidtable[$v] ) ) {
1428                                 foreach ( $vmarked as $vo ) {
1429                                         // use syntax: -{A|zh:WordZh;zh-tw:WordTw}-
1430                                         // or -{H|zh:WordZh;zh-tw:WordTw}-
1431                                         // or -{-|zh:WordZh;zh-tw:WordTw}-
1432                                         // to introduce a custom mapping between
1433                                         // words WordZh and WordTw in the whole text
1434                                         if ( $manLevel[$v] == 'bidirectional' ) {
1435                                                 $this->mConvTable[$v][$bidtable[$vo]] = $bidtable[$v];
1436                                         }
1437                                         if ( $manLevel[$vo] == 'bidirectional' ) {
1438                                                 $this->mConvTable[$vo][$bidtable[$v]] = $bidtable[$vo];
1439                                         }
1440                                 }
1441                                 $vmarked[] = $v;
1442                         }
1443                         /* for unidirectional array fill to convert tables */
1444                         if ( ( $manLevel[$v] == 'bidirectional' || $manLevel[$v] == 'unidirectional' )
1445                                 && isset( $unidtable[$v] ) )
1446                         {
1447                                 if ( isset( $this->mConvTable[$v] ) ) {
1448                                         $this->mConvTable[$v] = array_merge( $this->mConvTable[$v], $unidtable[$v] );
1449                                 } else {
1450                                         $this->mConvTable[$v] = $unidtable[$v];
1451                                 }
1452                         }
1453                 }
1454         }
1455
1456         /**
1457          * Parse rules and flags.
1458          * @param $variant String: variant language code
1459          */
1460         public function parse( $variant = null ) {
1461                 if ( !$variant ) {
1462                         $variant = $this->mConverter->getPreferredVariant();
1463                 }
1464
1465                 $this->parseFlags();
1466                 $flags = $this->mFlags;
1467
1468                 // convert to specified variant
1469                 // syntax: -{zh-hans;zh-hant[;...]|<text to convert>}-
1470                 if ( $this->mVariantFlags ) {
1471                         // check if current variant in flags
1472                         if ( isset( $this->mVariantFlags[$variant] ) ) {
1473                                 // then convert <text to convert> to current language
1474                                 $this->mRules = $this->mConverter->autoConvert( $this->mRules,
1475                                         $variant );
1476                         } else {
1477                                 // if current variant no in flags,
1478                                 // then we check its fallback variants.
1479                                 $variantFallbacks =
1480                                         $this->mConverter->getVariantFallbacks( $variant );
1481                                 if ( is_array( $variantFallbacks ) ) {
1482                                         foreach ( $variantFallbacks as $variantFallback ) {
1483                                                 // if current variant's fallback exist in flags
1484                                                 if ( isset( $this->mVariantFlags[$variantFallback] ) ) {
1485                                                         // then convert <text to convert> to fallback language
1486                                                         $this->mRules =
1487                                                                 $this->mConverter->autoConvert( $this->mRules,
1488                                                                         $variantFallback );
1489                                                         break;
1490                                                 }
1491                                         }
1492                                 }
1493                         }
1494                         $this->mFlags = $flags = array( 'R' => true );
1495                 }
1496
1497                 if ( !isset( $flags['R'] ) && !isset( $flags['N'] ) ) {
1498                         // decode => HTML entities modified by Sanitizer::removeHTMLtags
1499                         $this->mRules = str_replace( '=&gt;', '=>', $this->mRules );
1500                         $this->parseRules();
1501                 }
1502                 $rules = $this->mRules;
1503
1504                 if ( !$this->mBidtable && !$this->mUnidtable ) {
1505                         if ( isset( $flags['+'] ) || isset( $flags['-'] ) ) {
1506                                 // fill all variants if text in -{A/H/-|text} without rules
1507                                 foreach ( $this->mConverter->mVariants as $v ) {
1508                                         $this->mBidtable[$v] = $rules;
1509                                 }
1510                         } elseif ( !isset( $flags['N'] ) && !isset( $flags['T'] ) ) {
1511                                 $this->mFlags = $flags = array( 'R' => true );
1512                         }
1513                 }
1514
1515                 $this->mRuleDisplay = false;
1516                 foreach ( $flags as $flag => $unused ) {
1517                         switch ( $flag ) {
1518                                 case 'R':
1519                                         // if we don't do content convert, still strip the -{}- tags
1520                                         $this->mRuleDisplay = $rules;
1521                                         break;
1522                                 case 'N':
1523                                         // process N flag: output current variant name
1524                                         $ruleVar = trim( $rules );
1525                                         if ( isset( $this->mConverter->mVariantNames[$ruleVar] ) ) {
1526                                                 $this->mRuleDisplay = $this->mConverter->mVariantNames[$ruleVar];
1527                                         } else {
1528                                                 $this->mRuleDisplay = '';
1529                                         }
1530                                         break;
1531                                 case 'D':
1532                                         // process D flag: output rules description
1533                                         $this->mRuleDisplay = $this->getRulesDesc();
1534                                         break;
1535                                 case 'H':
1536                                         // process H,- flag or T only: output nothing
1537                                         $this->mRuleDisplay = '';
1538                                         break;
1539                                 case '-':
1540                                         $this->mRulesAction = 'remove';
1541                                         $this->mRuleDisplay = '';
1542                                         break;
1543                                 case '+':
1544                                         $this->mRulesAction = 'add';
1545                                         $this->mRuleDisplay = '';
1546                                         break;
1547                                 case 'S':
1548                                         $this->mRuleDisplay = $this->getRuleConvertedStr( $variant );
1549                                         break;
1550                                 case 'T':
1551                                         $this->mRuleTitle = $this->getRuleConvertedTitle( $variant );
1552                                         $this->mRuleDisplay = '';
1553                                         break;
1554                                 default:
1555                                         // ignore unknown flags (but see error case below)
1556                         }
1557                 }
1558                 if ( $this->mRuleDisplay === false ) {
1559                         $this->mRuleDisplay = '<span class="error">'
1560                                 . wfMessage( 'converter-manual-rule-error' )->inContentLanguage()->escaped()
1561                                 . '</span>';
1562                 }
1563
1564                 $this->generateConvTable();
1565         }
1566
1567         /**
1568          * @todo FIXME: code this function :)
1569          */
1570         public function hasRules() {
1571                 // TODO:
1572         }
1573
1574         /**
1575          * Get display text on markup -{...}-
1576          * @return string
1577          */
1578         public function getDisplay() {
1579                 return $this->mRuleDisplay;
1580         }
1581
1582         /**
1583          * Get converted title.
1584          * @return string
1585          */
1586         public function getTitle() {
1587                 return $this->mRuleTitle;
1588         }
1589
1590         /**
1591          * Return how deal with conversion rules.
1592          * @return string
1593          */
1594         public function getRulesAction() {
1595                 return $this->mRulesAction;
1596         }
1597
1598         /**
1599          * Get conversion table. (bidirectional and unidirectional
1600          * conversion table)
1601          * @return array
1602          */
1603         public function getConvTable() {
1604                 return $this->mConvTable;
1605         }
1606
1607         /**
1608          * Get conversion rules string.
1609          * @return string
1610          */
1611         public function getRules() {
1612                 return $this->mRules;
1613         }
1614
1615         /**
1616          * Get conversion flags.
1617          * @return array
1618          */
1619         public function getFlags() {
1620                 return $this->mFlags;
1621         }
1622 }