languages/LanguageConverter.php

   1 <?php
   2 /**
   3  * This program is free software; you can redistribute it and/or modify
   4  * it under the terms of the GNU General Public License as published by
   5  * the Free Software Foundation; either version 2 of the License, or
   6  * (at your option) any later version.
   7  *
   8  * This program is distributed in the hope that it will be useful,
   9  * but WITHOUT ANY WARRANTY; without even the implied warranty of
  10  * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the
  11  * GNU General Public License for more details.
  12  *
  13  * You should have received a copy of the GNU General Public License along
  14  * with this program; if not, write to the Free Software Foundation, Inc.,
  15  * 51 Franklin Street, Fifth Floor, Boston, MA 02110-1301, USA.
  16  * http://www.gnu.org/copyleft/gpl.html
  17  *
  18  * @file
  19  * @ingroup Language
  20  */
  21
  22 /**
  23  * Base class for language conversion.
  24  * @ingroup Language
  25  *
  26  * @author Zhengzhu Feng <zhengzhu@gmail.com>
  27  * @author fdcn <fdcn64@gmail.com>
  28  * @author shinjiman <shinjiman@gmail.com>
  29  * @author PhiLiP <philip.npc@gmail.com>
  30  */
  31 class LanguageConverter {
  32         /**
  33          * languages supporting variants
  34          * @since 1.20
  35          * @var array
  36          */
  37         static public $languagesWithVariants = array(
  38                 'gan',
  39                 'iu',
  40                 'kk',
  41                 'ku',
  42                 'shi',
  43                 'sr',
  44                 'tg',
  45                 'uz',
  46                 'zh',
  47         );
  48
  49         public $mMainLanguageCode;
  50         public $mVariants, $mVariantFallbacks, $mVariantNames;
  51         public $mTablesLoaded = false;
  52         public $mTables;
  53         // 'bidirectional' 'unidirectional' 'disable' for each variant
  54         public $mManualLevel;
  55
  56         /**
  57          * @var string Memcached key name
  58          */
  59         public $mCacheKey;
  60
  61         public $mLangObj;
  62         public $mFlags;
  63         public $mDescCodeSep = ':', $mDescVarSep = ';';
  64         public $mUcfirst = false;
  65         public $mConvRuleTitle = false;
  66         public $mURLVariant;
  67         public $mUserVariant;
  68         public $mHeaderVariant;
  69         public $mMaxDepth = 10;
  70         public $mVarSeparatorPattern;
  71
  72         const CACHE_VERSION_KEY = 'VERSION 7';
  73
  74         /**
  75          * Constructor
  76          *
  77          * @param Language $langobj
  78          * @param string $maincode The main language code of this language
  79          * @param array $variants The supported variants of this language
  80          * @param array $variantfallbacks The fallback language of each variant
  81          * @param array $flags Defining the custom strings that maps to the flags
  82          * @param array $manualLevel Limit for supported variants
  83          */
  84         public function __construct( $langobj, $maincode, $variants = array(),
  85                                                                 $variantfallbacks = array(), $flags = array(),
  86                                                                 $manualLevel = array() ) {
  87                 global $wgDisabledVariants;
  88                 $this->mLangObj = $langobj;
  89                 $this->mMainLanguageCode = $maincode;
  90                 $this->mVariants = array_diff( $variants, $wgDisabledVariants );
  91                 $this->mVariantFallbacks = $variantfallbacks;
  92                 $this->mVariantNames = Language::fetchLanguageNames();
  93                 $this->mCacheKey = wfMemcKey( 'conversiontables', $maincode );
  94                 $defaultflags = array(
  95                         // 'S' show converted text
  96                         // '+' add rules for alltext
  97                         // 'E' the gave flags is error
  98                         // these flags above are reserved for program
  99                         'A' => 'A',       // add rule for convert code (all text convert)
 100                         'T' => 'T',       // title convert
 101                         'R' => 'R',       // raw content
 102                         'D' => 'D',       // convert description (subclass implement)
 103                         '-' => '-',       // remove convert (not implement)
 104                         'H' => 'H',       // add rule for convert code (but no display in placed code)
 105                         'N' => 'N'        // current variant name
 106                 );
 107                 $this->mFlags = array_merge( $defaultflags, $flags );
 108                 foreach ( $this->mVariants as $v ) {
 109                         if ( array_key_exists( $v, $manualLevel ) ) {
 110                                 $this->mManualLevel[$v] = $manualLevel[$v];
 111                         } else {
 112                                 $this->mManualLevel[$v] = 'bidirectional';
 113                         }
 114                         $this->mFlags[$v] = $v;
 115                 }
 116         }
 117
 118         /**
 119          * Get all valid variants.
 120          * Call this instead of using $this->mVariants directly.
 121          *
 122          * @return array Contains all valid variants
 123          */
 124         public function getVariants() {
 125                 return $this->mVariants;
 126         }
 127
 128         /**
 129          * In case some variant is not defined in the markup, we need
 130          * to have some fallback. For example, in zh, normally people
 131          * will define zh-hans and zh-hant, but less so for zh-sg or zh-hk.
 132          * when zh-sg is preferred but not defined, we will pick zh-hans
 133          * in this case. Right now this is only used by zh.
 134          *
 135          * @param string $variant The language code of the variant
 136          * @return string|array The code of the fallback language or the
 137          *   main code if there is no fallback
 138          */
 139         public function getVariantFallbacks( $variant ) {
 140                 if ( isset( $this->mVariantFallbacks[$variant] ) ) {
 141                         return $this->mVariantFallbacks[$variant];
 142                 }
 143                 return $this->mMainLanguageCode;
 144         }
 145
 146         /**
 147          * Get the title produced by the conversion rule.
 148          * @return string The converted title text
 149          */
 150         public function getConvRuleTitle() {
 151                 return $this->mConvRuleTitle;
 152         }
 153
 154         /**
 155          * Get preferred language variant.
 156          * @return string The preferred language code
 157          */
 158         public function getPreferredVariant() {
 159                 global $wgDefaultLanguageVariant, $wgUser;
 160
 161                 $req = $this->getURLVariant();
 162
 163                 if ( $wgUser->isLoggedIn() && !$req ) {
 164                         $req = $this->getUserVariant();
 165                 } elseif ( !$req ) {
 166                         $req = $this->getHeaderVariant();
 167                 }
 168
 169                 if ( $wgDefaultLanguageVariant && !$req ) {
 170                         $req = $this->validateVariant( $wgDefaultLanguageVariant );
 171                 }
 172
 173                 // This function, unlike the other get*Variant functions, is
 174                 // not memoized (i.e. there return value is not cached) since
 175                 // new information might appear during processing after this
 176                 // is first called.
 177                 if ( $this->validateVariant( $req ) ) {
 178                         return $req;
 179                 }
 180                 return $this->mMainLanguageCode;
 181         }
 182
 183         /**
 184          * Get default variant.
 185          * This function would not be affected by user's settings
 186          * @return string The default variant code
 187          */
 188         public function getDefaultVariant() {
 189                 global $wgDefaultLanguageVariant;
 190
 191                 $req = $this->getURLVariant();
 192
 193                 if ( !$req ) {
 194                         $req = $this->getHeaderVariant();
 195                 }
 196
 197                 if ( $wgDefaultLanguageVariant && !$req ) {
 198                         $req = $this->validateVariant( $wgDefaultLanguageVariant );
 199                 }
 200
 201                 if ( $req ) {
 202                         return $req;
 203                 }
 204                 return $this->mMainLanguageCode;
 205         }
 206
 207         /**
 208          * Validate the variant
 209          * @param string $variant The variant to validate
 210          * @return mixed Returns the variant if it is valid, null otherwise
 211          */
 212         public function validateVariant( $variant = null ) {
 213                 if ( $variant !== null && in_array( $variant, $this->mVariants ) ) {
 214                         return $variant;
 215                 }
 216                 return null;
 217         }
 218
 219         /**
 220          * Get the variant specified in the URL
 221          *
 222          * @return mixed Variant if one found, false otherwise.
 223          */
 224         public function getURLVariant() {
 225                 global $wgRequest;
 226
 227                 if ( $this->mURLVariant ) {
 228                         return $this->mURLVariant;
 229                 }
 230
 231                 // see if the preference is set in the request
 232                 $ret = $wgRequest->getText( 'variant' );
 233
 234                 if ( !$ret ) {
 235                         $ret = $wgRequest->getVal( 'uselang' );
 236                 }
 237
 238                 $this->mURLVariant = $this->validateVariant( $ret );
 239                 return $this->mURLVariant;
 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                 $this->mUserVariant = $this->validateVariant( $ret );
 273                 return $this->mUserVariant;
 274         }
 275
 276         /**
 277          * Determine the language variant from the Accept-Language header.
 278          *
 279          * @return mixed Variant if one found, false otherwise.
 280          */
 281         protected function getHeaderVariant() {
 282                 global $wgRequest;
 283
 284                 if ( $this->mHeaderVariant ) {
 285                         return $this->mHeaderVariant;
 286                 }
 287
 288                 // see if some supported language variant is set in the
 289                 // HTTP header.
 290                 $languages = array_keys( $wgRequest->getAcceptLang() );
 291                 if ( empty( $languages ) ) {
 292                         return null;
 293                 }
 294
 295                 $fallbackLanguages = array();
 296                 foreach ( $languages as $language ) {
 297                         $this->mHeaderVariant = $this->validateVariant( $language );
 298                         if ( $this->mHeaderVariant ) {
 299                                 break;
 300                         }
 301
 302                         // To see if there are fallbacks of current language.
 303                         // We record these fallback variants, and process
 304                         // them later.
 305                         $fallbacks = $this->getVariantFallbacks( $language );
 306                         if ( is_string( $fallbacks ) && $fallbacks !== $this->mMainLanguageCode ) {
 307                                 $fallbackLanguages[] = $fallbacks;
 308                         } elseif ( is_array( $fallbacks ) ) {
 309                                 $fallbackLanguages =
 310                                         array_merge( $fallbackLanguages, $fallbacks );
 311                         }
 312                 }
 313
 314                 if ( !$this->mHeaderVariant ) {
 315                         // process fallback languages now
 316                         $fallback_languages = array_unique( $fallbackLanguages );
 317                         foreach ( $fallback_languages as $language ) {
 318                                 $this->mHeaderVariant = $this->validateVariant( $language );
 319                                 if ( $this->mHeaderVariant ) {
 320                                         break;
 321                                 }
 322                         }
 323                 }
 324
 325                 return $this->mHeaderVariant;
 326         }
 327
 328         /**
 329          * Dictionary-based conversion.
 330          * This function would not parse the conversion rules.
 331          * If you want to parse rules, try to use convert() or
 332          * convertTo().
 333          *
 334          * @param string $text The text to be converted
 335          * @param bool|string $toVariant The target language code
 336          * @return string The converted text
 337          */
 338         public function autoConvert( $text, $toVariant = false ) {
 339
 340                 $this->loadTables();
 341
 342                 if ( !$toVariant ) {
 343                         $toVariant = $this->getPreferredVariant();
 344                         if ( !$toVariant ) {
 345                                 return $text;
 346                         }
 347                 }
 348
 349                 if ( $this->guessVariant( $text, $toVariant ) ) {
 350                         return $text;
 351                 }
 352
 353                 /* we convert everything except:
 354                    1. HTML markups (anything between < and >)
 355                    2. HTML entities
 356                    3. placeholders created by the parser
 357                 */
 358                 $marker = '|' . Parser::MARKER_PREFIX . '[\-a-zA-Z0-9]+';
 359
 360                 // this one is needed when the text is inside an HTML markup
 361                 $htmlfix = '|<[^>]+$|^[^<>]*>';
 362
 363                 // disable convert to variants between <code> tags
 364                 $codefix = '<code>.+?<\/code>|';
 365                 // disable conversion of <script> tags
 366                 $scriptfix = '<script.*?>.*?<\/script>|';
 367                 // disable conversion of <pre> tags
 368                 $prefix = '<pre.*?>.*?<\/pre>|';
 369
 370                 $reg = '/' . $codefix . $scriptfix . $prefix .
 371                         '<[^>]+>|&[a-zA-Z#][a-z0-9]+;' . $marker . $htmlfix . '/s';
 372                 $startPos = 0;
 373                 $sourceBlob = '';
 374                 $literalBlob = '';
 375
 376                 // Guard against delimiter nulls in the input
 377                 $text = str_replace( "\000", '', $text );
 378
 379                 $markupMatches = null;
 380                 $elementMatches = null;
 381                 while ( $startPos < strlen( $text ) ) {
 382                         if ( preg_match( $reg, $text, $markupMatches, PREG_OFFSET_CAPTURE, $startPos ) ) {
 383                                 $elementPos = $markupMatches[0][1];
 384                                 $element = $markupMatches[0][0];
 385                         } else {
 386                                 $elementPos = strlen( $text );
 387                                 $element = '';
 388                         }
 389
 390                         // Queue the part before the markup for translation in a batch
 391                         $sourceBlob .= substr( $text, $startPos, $elementPos - $startPos ) . "\000";
 392
 393                         // Advance to the next position
 394                         $startPos = $elementPos + strlen( $element );
 395
 396                         // Translate any alt or title attributes inside the matched element
 397                         if ( $element !== ''
 398                                 && preg_match( '/^(<[^>\s]*)\s([^>]*)(.*)$/', $element, $elementMatches )
 399                         ) {
 400                                 $attrs = Sanitizer::decodeTagAttributes( $elementMatches[2] );
 401                                 $changed = false;
 402                                 foreach ( array( 'title', 'alt' ) as $attrName ) {
 403                                         if ( !isset( $attrs[$attrName] ) ) {
 404                                                 continue;
 405                                         }
 406                                         $attr = $attrs[$attrName];
 407                                         // Don't convert URLs
 408                                         if ( !strpos( $attr, '://' ) ) {
 409                                                 $attr = $this->recursiveConvertTopLevel( $attr, $toVariant );
 410                                         }
 411
 412                                         // Remove HTML tags to avoid disrupting the layout
 413                                         $attr = preg_replace( '/<[^>]+>/', '', $attr );
 414                                         if ( $attr !== $attrs[$attrName] ) {
 415                                                 $attrs[$attrName] = $attr;
 416                                                 $changed = true;
 417                                         }
 418                                 }
 419                                 if ( $changed ) {
 420                                         $element = $elementMatches[1] . Html::expandAttributes( $attrs ) .
 421                                                 $elementMatches[3];
 422                                 }
 423                         }
 424                         $literalBlob .= $element . "\000";
 425                 }
 426
 427                 // Do the main translation batch
 428                 $translatedBlob = $this->translate( $sourceBlob, $toVariant );
 429
 430                 // Put the output back together
 431                 $translatedIter = StringUtils::explode( "\000", $translatedBlob );
 432                 $literalIter = StringUtils::explode( "\000", $literalBlob );
 433                 $output = '';
 434                 while ( $translatedIter->valid() && $literalIter->valid() ) {
 435                         $output .= $translatedIter->current();
 436                         $output .= $literalIter->current();
 437                         $translatedIter->next();
 438                         $literalIter->next();
 439                 }
 440
 441                 return $output;
 442         }
 443
 444         /**
 445          * Translate a string to a variant.
 446          * Doesn't parse rules or do any of that other stuff, for that use
 447          * convert() or convertTo().
 448          *
 449          * @param string $text Text to convert
 450          * @param string $variant Variant language code
 451          * @return string Translated text
 452          */
 453         public function translate( $text, $variant ) {
 454                 // If $text is empty or only includes spaces, do nothing
 455                 // Otherwise translate it
 456                 if ( trim( $text ) ) {
 457                         $this->loadTables();
 458                         $text = $this->mTables[$variant]->replace( $text );
 459                 }
 460                 return $text;
 461         }
 462
 463         /**
 464          * Call translate() to convert text to all valid variants.
 465          *
 466          * @param string $text The text to be converted
 467          * @return array Variant => converted text
 468          */
 469         public function autoConvertToAllVariants( $text ) {
 470                 $this->loadTables();
 471
 472                 $ret = array();
 473                 foreach ( $this->mVariants as $variant ) {
 474                         $ret[$variant] = $this->translate( $text, $variant );
 475                 }
 476
 477                 return $ret;
 478         }
 479
 480         /**
 481          * Apply manual conversion rules.
 482          *
 483          * @param ConverterRule $convRule
 484          */
 485         protected function applyManualConv( $convRule ) {
 486                 // Use syntax -{T|zh-cn:TitleCN; zh-tw:TitleTw}- to custom
 487                 // title conversion.
 488                 // Bug 24072: $mConvRuleTitle was overwritten by other manual
 489                 // rule(s) not for title, this breaks the title conversion.
 490                 $newConvRuleTitle = $convRule->getTitle();
 491                 if ( $newConvRuleTitle ) {
 492                         // So I add an empty check for getTitle()
 493                         $this->mConvRuleTitle = $newConvRuleTitle;
 494                 }
 495
 496                 // merge/remove manual conversion rules to/from global table
 497                 $convTable = $convRule->getConvTable();
 498                 $action = $convRule->getRulesAction();
 499                 foreach ( $convTable as $variant => $pair ) {
 500                         if ( !$this->validateVariant( $variant ) ) {
 501                                 continue;
 502                         }
 503
 504                         if ( $action == 'add' ) {
 505                                 // More efficient than array_merge(), about 2.5 times.
 506                                 foreach ( $pair as $from => $to ) {
 507                                         $this->mTables[$variant]->setPair( $from, $to );
 508                                 }
 509                         } elseif ( $action == 'remove' ) {
 510                                 $this->mTables[$variant]->removeArray( $pair );
 511                         }
 512                 }
 513         }
 514
 515         /**
 516          * Auto convert a Title object to a readable string in the
 517          * preferred variant.
 518          *
 519          * @param Title $title A object of Title
 520          * @return string Converted title text
 521          */
 522         public function convertTitle( $title ) {
 523                 $variant = $this->getPreferredVariant();
 524                 $index = $title->getNamespace();
 525                 if ( $index !== NS_MAIN ) {
 526                         $text = $this->convertNamespace( $index, $variant ) . ':';
 527                 } else {
 528                         $text = '';
 529                 }
 530                 $text .= $this->translate( $title->getText(), $variant );
 531                 return $text;
 532         }
 533
 534         /**
 535          * Get the namespace display name in the preferred variant.
 536          *
 537          * @param int $index Namespace id
 538          * @param string|null $variant Variant code or null for preferred variant
 539          * @return string Namespace name for display
 540          */
 541         public function convertNamespace( $index, $variant = null ) {
 542                 if ( $index === NS_MAIN ) {
 543                         return '';
 544                 }
 545
 546                 if ( $variant === null ) {
 547                         $variant = $this->getPreferredVariant();
 548                 }
 549
 550                 $cache = ObjectCache::newAccelerator( CACHE_NONE );
 551                 $key = wfMemcKey( 'languageconverter', 'namespace-text', $index, $variant );
 552                 $nsVariantText = $cache->get( $key );
 553                 if ( $nsVariantText !== false ) {
 554                         return $nsVariantText;
 555                 }
 556
 557                 // First check if a message gives a converted name in the target variant.
 558                 $nsConvMsg = wfMessage( 'conversion-ns' . $index )->inLanguage( $variant );
 559                 if ( $nsConvMsg->exists() ) {
 560                         $nsVariantText = $nsConvMsg->plain();
 561                 }
 562
 563                 // Then check if a message gives a converted name in content language
 564                 // which needs extra translation to the target variant.
 565                 if ( $nsVariantText === false ) {
 566                         $nsConvMsg = wfMessage( 'conversion-ns' . $index )->inContentLanguage();
 567                         if ( $nsConvMsg->exists() ) {
 568                                 $nsVariantText = $this->translate( $nsConvMsg->plain(), $variant );
 569                         }
 570                 }
 571
 572                 if ( $nsVariantText === false ) {
 573                         // No message exists, retrieve it from the target variant's namespace names.
 574                         $langObj = $this->mLangObj->factory( $variant );
 575                         $nsVariantText = $langObj->getFormattedNsText( $index );
 576                 }
 577
 578                 $cache->set( $key, $nsVariantText, 60 );
 579
 580                 return $nsVariantText;
 581         }
 582
 583         /**
 584          * Convert text to different variants of a language. The automatic
 585          * conversion is done in autoConvert(). Here we parse the text
 586          * marked with -{}-, which specifies special conversions of the
 587          * text that can not be accomplished in autoConvert().
 588          *
 589          * Syntax of the markup:
 590          * -{code1:text1;code2:text2;...}-  or
 591          * -{flags|code1:text1;code2:text2;...}-  or
 592          * -{text}- in which case no conversion should take place for text
 593          *
 594          * @param string $text Text to be converted
 595          * @return string Converted text
 596          */
 597         public function convert( $text ) {
 598                 $variant = $this->getPreferredVariant();
 599                 return $this->convertTo( $text, $variant );
 600         }
 601
 602         /**
 603          * Same as convert() except a extra parameter to custom variant.
 604          *
 605          * @param string $text Text to be converted
 606          * @param string $variant The target variant code
 607          * @return string Converted text
 608          */
 609         public function convertTo( $text, $variant ) {
 610                 global $wgDisableLangConversion;
 611                 if ( $wgDisableLangConversion ) {
 612                         return $text;
 613                 }
 614                 // Reset converter state for a new converter run.
 615                 $this->mConvRuleTitle = false;
 616                 return $this->recursiveConvertTopLevel( $text, $variant );
 617         }
 618
 619         /**
 620          * Recursively convert text on the outside. Allow to use nested
 621          * markups to custom rules.
 622          *
 623          * @param string $text Text to be converted
 624          * @param string $variant The target variant code
 625          * @param int $depth Depth of recursion
 626          * @return string Converted text
 627          */
 628         protected function recursiveConvertTopLevel( $text, $variant, $depth = 0 ) {
 629                 $startPos = 0;
 630                 $out = '';
 631                 $length = strlen( $text );
 632                 $shouldConvert = !$this->guessVariant( $text, $variant );
 633
 634                 while ( $startPos < $length ) {
 635                         $pos = strpos( $text, '-{', $startPos );
 636
 637                         if ( $pos === false ) {
 638                                 // No more markup, append final segment
 639                                 $fragment = substr( $text, $startPos );
 640                                 $out .= $shouldConvert ? $this->autoConvert( $fragment, $variant ) : $fragment;
 641                                 return $out;
 642                         }
 643
 644                         // Markup found
 645                         // Append initial segment
 646                         $fragment = substr( $text, $startPos, $pos - $startPos );
 647                         $out .= $shouldConvert ? $this->autoConvert( $fragment, $variant ) : $fragment;
 648
 649                         // Advance position
 650                         $startPos = $pos;
 651
 652                         // Do recursive conversion
 653                         $out .= $this->recursiveConvertRule( $text, $variant, $startPos, $depth + 1 );
 654                 }
 655
 656                 return $out;
 657         }
 658
 659         /**
 660          * Recursively convert text on the inside.
 661          *
 662          * @param string $text Text to be converted
 663          * @param string $variant The target variant code
 664          * @param int $startPos
 665          * @param int $depth Depth of recursion
 666          *
 667          * @throws MWException
 668          * @return string Converted text
 669          */
 670         protected function recursiveConvertRule( $text, $variant, &$startPos, $depth = 0 ) {
 671                 // Quick sanity check (no function calls)
 672                 if ( $text[$startPos] !== '-' || $text[$startPos + 1] !== '{' ) {
 673                         throw new MWException( __METHOD__ . ': invalid input string' );
 674                 }
 675
 676                 $startPos += 2;
 677                 $inner = '';
 678                 $warningDone = false;
 679                 $length = strlen( $text );
 680
 681                 while ( $startPos < $length ) {
 682                         $m = false;
 683                         preg_match( '/-\{|\}-/', $text, $m, PREG_OFFSET_CAPTURE, $startPos );
 684                         if ( !$m ) {
 685                                 // Unclosed rule
 686                                 break;
 687                         }
 688
 689                         $token = $m[0][0];
 690                         $pos = $m[0][1];
 691
 692                         // Markup found
 693                         // Append initial segment
 694                         $inner .= substr( $text, $startPos, $pos - $startPos );
 695
 696                         // Advance position
 697                         $startPos = $pos;
 698
 699                         switch ( $token ) {
 700                                 case '-{':
 701                                         // Check max depth
 702                                         if ( $depth >= $this->mMaxDepth ) {
 703                                                 $inner .= '-{';
 704                                                 if ( !$warningDone ) {
 705                                                         $inner .= '<span class="error">' .
 706                                                                 wfMessage( 'language-converter-depth-warning' )
 707                                                                         ->numParams( $this->mMaxDepth )->inContentLanguage()->text() .
 708                                                                 '</span>';
 709                                                         $warningDone = true;
 710                                                 }
 711                                                 $startPos += 2;
 712                                                 continue;
 713                                         }
 714                                         // Recursively parse another rule
 715                                         $inner .= $this->recursiveConvertRule( $text, $variant, $startPos, $depth + 1 );
 716                                         break;
 717                                 case '}-':
 718                                         // Apply the rule
 719                                         $startPos += 2;
 720                                         $rule = new ConverterRule( $inner, $this );
 721                                         $rule->parse( $variant );
 722                                         $this->applyManualConv( $rule );
 723                                         return $rule->getDisplay();
 724                                 default:
 725                                         throw new MWException( __METHOD__ . ': invalid regex match' );
 726                         }
 727                 }
 728
 729                 // Unclosed rule
 730                 if ( $startPos < $length ) {
 731                         $inner .= substr( $text, $startPos );
 732                 }
 733                 $startPos = $length;
 734                 return '-{' . $this->autoConvert( $inner, $variant );
 735         }
 736
 737         /**
 738          * If a language supports multiple variants, it is possible that
 739          * non-existing link in one variant actually exists in another variant.
 740          * This function tries to find it. See e.g. LanguageZh.php
 741          * The input parameters may be modified upon return
 742          *
 743          * @param string &$link The name of the link
 744          * @param Title &$nt The title object of the link
 745          * @param bool $ignoreOtherCond To disable other conditions when
 746          *   we need to transclude a template or update a category's link
 747          */
 748         public function findVariantLink( &$link, &$nt, $ignoreOtherCond = false ) {
 749                 # If the article has already existed, there is no need to
 750                 # check it again, otherwise it may cause a fault.
 751                 if ( is_object( $nt ) && $nt->exists() ) {
 752                         return;
 753                 }
 754
 755                 global $wgDisableLangConversion, $wgDisableTitleConversion, $wgRequest;
 756                 $isredir = $wgRequest->getText( 'redirect', 'yes' );
 757                 $action = $wgRequest->getText( 'action' );
 758                 if ( $action == 'edit' && $wgRequest->getBool( 'redlink' ) ) {
 759                         $action = 'view';
 760                 }
 761                 $linkconvert = $wgRequest->getText( 'linkconvert', 'yes' );
 762                 $disableLinkConversion = $wgDisableLangConversion
 763                         || $wgDisableTitleConversion;
 764                 $linkBatch = new LinkBatch();
 765
 766                 $ns = NS_MAIN;
 767
 768                 if ( $disableLinkConversion ||
 769                         ( !$ignoreOtherCond &&
 770                                 ( $isredir == 'no'
 771                                         || $action == 'edit'
 772                                         || $action == 'submit'
 773                                         || $linkconvert == 'no' ) ) ) {
 774                         return;
 775                 }
 776
 777                 if ( is_object( $nt ) ) {
 778                         $ns = $nt->getNamespace();
 779                 }
 780
 781                 $variants = $this->autoConvertToAllVariants( $link );
 782                 if ( !$variants ) { // give up
 783                         return;
 784                 }
 785
 786                 $titles = array();
 787
 788                 foreach ( $variants as $v ) {
 789                         if ( $v != $link ) {
 790                                 $varnt = Title::newFromText( $v, $ns );
 791                                 if ( !is_null( $varnt ) ) {
 792                                         $linkBatch->addObj( $varnt );
 793                                         $titles[] = $varnt;
 794                                 }
 795                         }
 796                 }
 797
 798                 // fetch all variants in single query
 799                 $linkBatch->execute();
 800
 801                 foreach ( $titles as $varnt ) {
 802                         if ( $varnt->getArticleID() > 0 ) {
 803                                 $nt = $varnt;
 804                                 $link = $varnt->getText();
 805                                 break;
 806                         }
 807                 }
 808         }
 809
 810         /**
 811          * Returns language specific hash options.
 812          *
 813          * @return string
 814          */
 815         public function getExtraHashOptions() {
 816                 $variant = $this->getPreferredVariant();
 817
 818                 return '!' . $variant;
 819         }
 820
 821         /**
 822          * Guess if a text is written in a variant. This should be implemented in subclasses.
 823          *
 824          * @param string $text The text to be checked
 825          * @param string $variant Language code of the variant to be checked for
 826          * @return bool True if $text appears to be written in $variant, false if not
 827          *
 828          * @author Nikola Smolenski <smolensk@eunet.rs>
 829          * @since 1.19
 830          */
 831         public function guessVariant( $text, $variant ) {
 832                 return false;
 833         }
 834
 835         /**
 836          * Load default conversion tables.
 837          * This method must be implemented in derived class.
 838          *
 839          * @private
 840          * @throws MWException
 841          */
 842         function loadDefaultTables() {
 843                 $name = get_class( $this );
 844
 845                 throw new MWException( "Must implement loadDefaultTables() method in class $name" );
 846         }
 847
 848         /**
 849          * Load conversion tables either from the cache or the disk.
 850          * @private
 851          * @param bool $fromCache Load from memcached? Defaults to true.
 852          */
 853         function loadTables( $fromCache = true ) {
 854                 global $wgLanguageConverterCacheType;
 855
 856                 if ( $this->mTablesLoaded ) {
 857                         return;
 858                 }
 859
 860                 $this->mTablesLoaded = true;
 861                 $this->mTables = false;
 862                 $cache = ObjectCache::getInstance( $wgLanguageConverterCacheType );
 863                 if ( $fromCache ) {
 864                         wfProfileIn( __METHOD__ . '-cache' );
 865                         $this->mTables = $cache->get( $this->mCacheKey );
 866                         wfProfileOut( __METHOD__ . '-cache' );
 867                 }
 868                 if ( !$this->mTables || !array_key_exists( self::CACHE_VERSION_KEY, $this->mTables ) ) {
 869                         wfProfileIn( __METHOD__ . '-recache' );
 870                         // not in cache, or we need a fresh reload.
 871                         // We will first load the default tables
 872                         // then update them using things in MediaWiki:Conversiontable/*
 873                         $this->loadDefaultTables();
 874                         foreach ( $this->mVariants as $var ) {
 875                                 $cached = $this->parseCachedTable( $var );
 876                                 $this->mTables[$var]->mergeArray( $cached );
 877                         }
 878
 879                         $this->postLoadTables();
 880                         $this->mTables[self::CACHE_VERSION_KEY] = true;
 881
 882                         $cache->set( $this->mCacheKey, $this->mTables, 43200 );
 883                         wfProfileOut( __METHOD__ . '-recache' );
 884                 }
 885         }
 886
 887         /**
 888          * Hook for post processing after conversion tables are loaded.
 889          */
 890         function postLoadTables() {
 891         }
 892
 893         /**
 894          * Reload the conversion tables.
 895          *
 896          * @private
 897          */
 898         function reloadTables() {
 899                 if ( $this->mTables ) {
 900                         unset( $this->mTables );
 901                 }
 902
 903                 $this->mTablesLoaded = false;
 904                 $this->loadTables( false );
 905         }
 906
 907         /**
 908          * Parse the conversion table stored in the cache.
 909          *
 910          * The tables should be in blocks of the following form:
 911          *              -{
 912          *                      word => word ;
 913          *                      word => word ;
 914          *                      ...
 915          *              }-
 916          *
 917          * To make the tables more manageable, subpages are allowed
 918          * and will be parsed recursively if $recursive == true.
 919          *
 920          * @param string $code Language code
 921          * @param string $subpage Subpage name
 922          * @param bool $recursive Parse subpages recursively? Defaults to true.
 923          *
 924          * @return array
 925          */
 926         function parseCachedTable( $code, $subpage = '', $recursive = true ) {
 927                 static $parsed = array();
 928
 929                 $key = 'Conversiontable/' . $code;
 930                 if ( $subpage ) {
 931                         $key .= '/' . $subpage;
 932                 }
 933                 if ( array_key_exists( $key, $parsed ) ) {
 934                         return array();
 935                 }
 936
 937                 $parsed[$key] = true;
 938
 939                 if ( $subpage === '' ) {
 940                         $txt = MessageCache::singleton()->getMsgFromNamespace( $key, $code );
 941                 } else {
 942                         $txt = false;
 943                         $title = Title::makeTitleSafe( NS_MEDIAWIKI, $key );
 944                         if ( $title && $title->exists() ) {
 945                                 $revision = Revision::newFromTitle( $title );
 946                                 if ( $revision ) {
 947                                         if ( $revision->getContentModel() == CONTENT_MODEL_WIKITEXT ) {
 948                                                 $txt = $revision->getContent( Revision::RAW )->getNativeData();
 949                                         }
 950
 951                                         // @todo in the future, use a specialized content model, perhaps based on json!
 952                                 }
 953                         }
 954                 }
 955
 956                 # Nothing to parse if there's no text
 957                 if ( $txt === false || $txt === null || $txt === '' ) {
 958                         return array();
 959                 }
 960
 961                 // get all subpage links of the form
 962                 // [[MediaWiki:Conversiontable/zh-xx/...|...]]
 963                 $linkhead = $this->mLangObj->getNsText( NS_MEDIAWIKI ) .
 964                         ':Conversiontable';
 965                 $subs = StringUtils::explode( '[[', $txt );
 966                 $sublinks = array();
 967                 foreach ( $subs as $sub ) {
 968                         $link = explode( ']]', $sub, 2 );
 969                         if ( count( $link ) != 2 ) {
 970                                 continue;
 971                         }
 972                         $b = explode( '|', $link[0], 2 );
 973                         $b = explode( '/', trim( $b[0] ), 3 );
 974                         if ( count( $b ) == 3 ) {
 975                                 $sublink = $b[2];
 976                         } else {
 977                                 $sublink = '';
 978                         }
 979
 980                         if ( $b[0] == $linkhead && $b[1] == $code ) {
 981                                 $sublinks[] = $sublink;
 982                         }
 983                 }
 984
 985                 // parse the mappings in this page
 986                 $blocks = StringUtils::explode( '-{', $txt );
 987                 $ret = array();
 988                 $first = true;
 989                 foreach ( $blocks as $block ) {
 990                         if ( $first ) {
 991                                 // Skip the part before the first -{
 992                                 $first = false;
 993                                 continue;
 994                         }
 995                         $mappings = explode( '}-', $block, 2 );
 996                         $stripped = str_replace( array( "'", '"', '*', '#' ), '', $mappings[0] );
 997                         $table = StringUtils::explode( ';', $stripped );
 998                         foreach ( $table as $t ) {
 999                                 $m = explode( '=>', $t, 3 );
1000                                 if ( count( $m ) != 2 ) {
1001                                         continue;
1002                                 }
1003                                 // trim any trailling comments starting with '//'
1004                                 $tt = explode( '//', $m[1], 2 );
1005                                 $ret[trim( $m[0] )] = trim( $tt[0] );
1006                         }
1007                 }
1008
1009                 // recursively parse the subpages
1010                 if ( $recursive ) {
1011                         foreach ( $sublinks as $link ) {
1012                                 $s = $this->parseCachedTable( $code, $link, $recursive );
1013                                 $ret = $s + $ret;
1014                         }
1015                 }
1016
1017                 if ( $this->mUcfirst ) {
1018                         foreach ( $ret as $k => $v ) {
1019                                 $ret[$this->mLangObj->ucfirst( $k )] = $this->mLangObj->ucfirst( $v );
1020                         }
1021                 }
1022                 return $ret;
1023         }
1024
1025         /**
1026          * Enclose a string with the "no conversion" tag. This is used by
1027          * various functions in the Parser.
1028          *
1029          * @param string $text Text to be tagged for no conversion
1030          * @param bool $noParse Unused
1031          * @return string The tagged text
1032          */
1033         public function markNoConversion( $text, $noParse = false ) {
1034                 # don't mark if already marked
1035                 if ( strpos( $text, '-{' ) || strpos( $text, '}-' ) ) {
1036                         return $text;
1037                 }
1038
1039                 $ret = "-{R|$text}-";
1040                 return $ret;
1041         }
1042
1043         /**
1044          * Convert the sorting key for category links. This should make different
1045          * keys that are variants of each other map to the same key.
1046          *
1047          * @param string $key
1048          *
1049          * @return string
1050          */
1051         function convertCategoryKey( $key ) {
1052                 return $key;
1053         }
1054
1055         /**
1056          * Refresh the cache of conversion tables when
1057          * MediaWiki:Conversiontable* is updated.
1058          *
1059          * @param Title $titleobj The Title of the page being updated
1060          */
1061         public function updateConversionTable( Title $titleobj ) {
1062                 if ( $titleobj->getNamespace() == NS_MEDIAWIKI ) {
1063                         $title = $titleobj->getDBkey();
1064                         $t = explode( '/', $title, 3 );
1065                         $c = count( $t );
1066                         if ( $c > 1 && $t[0] == 'Conversiontable' ) {
1067                                 if ( $this->validateVariant( $t[1] ) ) {
1068                                         $this->reloadTables();
1069                                 }
1070                         }
1071                 }
1072         }
1073
1074         /**
1075          * Get the cached separator pattern for ConverterRule::parseRules()
1076          * @return string
1077          */
1078         function getVarSeparatorPattern() {
1079                 if ( is_null( $this->mVarSeparatorPattern ) ) {
1080                         // varsep_pattern for preg_split:
1081                         // text should be splited by ";" only if a valid variant
1082                         // name exist after the markup, for example:
1083                         //  -{zh-hans:<span style="font-size:120%;">xxx</span>;zh-hant:\
1084                         //      <span style="font-size:120%;">yyy</span>;}-
1085                         // we should split it as:
1086                         //  array(
1087                         //        [0] => 'zh-hans:<span style="font-size:120%;">xxx</span>'
1088                         //        [1] => 'zh-hant:<span style="font-size:120%;">yyy</span>'
1089                         //        [2] => ''
1090                         //       )
1091                         $pat = '/;\s*(?=';
1092                         foreach ( $this->mVariants as $variant ) {
1093                                 // zh-hans:xxx;zh-hant:yyy
1094                                 $pat .= $variant . '\s*:|';
1095                                 // xxx=>zh-hans:yyy; xxx=>zh-hant:zzz
1096                                 $pat .= '[^;]*?=>\s*' . $variant . '\s*:|';
1097                         }
1098                         $pat .= '\s*$)/';
1099                         $this->mVarSeparatorPattern = $pat;
1100                 }
1101                 return $this->mVarSeparatorPattern;
1102         }
1103 }