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 use MediaWiki\MediaWikiServices;
  22
  23 /**
  24  * Base class for language conversion.
  25  * @ingroup Language
  26  *
  27  * @author Zhengzhu Feng <zhengzhu@gmail.com>
  28  * @author fdcn <fdcn64@gmail.com>
  29  * @author shinjiman <shinjiman@gmail.com>
  30  * @author PhiLiP <philip.npc@gmail.com>
  31  */
  32 class LanguageConverter {
  33         /**
  34          * languages supporting variants
  35          * @since 1.20
  36          * @var array
  37          */
  38         static public $languagesWithVariants = [
  39                 'en',
  40                 'gan',
  41                 'iu',
  42                 'kk',
  43                 'ku',
  44                 'shi',
  45                 'sr',
  46                 'tg',
  47                 'uz',
  48                 'zh',
  49         ];
  50
  51         public $mMainLanguageCode;
  52
  53         /**
  54          * @var string[]
  55          */
  56         public $mVariants;
  57         public $mVariantFallbacks;
  58         public $mVariantNames;
  59         public $mTablesLoaded = false;
  60         public $mTables;
  61         // 'bidirectional' 'unidirectional' 'disable' for each variant
  62         public $mManualLevel;
  63
  64         public $mLangObj;
  65         public $mFlags;
  66         public $mDescCodeSep = ':', $mDescVarSep = ';';
  67         public $mUcfirst = false;
  68         public $mConvRuleTitle = false;
  69         public $mURLVariant;
  70         public $mUserVariant;
  71         public $mHeaderVariant;
  72         public $mMaxDepth = 10;
  73         public $mVarSeparatorPattern;
  74
  75         const CACHE_VERSION_KEY = 'VERSION 7';
  76
  77         /**
  78          * @param Language $langobj
  79          * @param string $maincode The main language code of this language
  80          * @param string[] $variants The supported variants of this language
  81          * @param array $variantfallbacks The fallback language of each variant
  82          * @param array $flags Defining the custom strings that maps to the flags
  83          * @param array $manualLevel Limit for supported variants
  84          */
  85         public function __construct( $langobj, $maincode, $variants = [],
  86                                                                 $variantfallbacks = [], $flags = [],
  87                                                                 $manualLevel = [] ) {
  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                 $defaultflags = [
  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 string[] 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 = $this->getURLVariant();
 162
 163                 Hooks::run( 'GetLangPreferredVariant', [ &$req ] );
 164
 165                 if ( $wgUser->isSafeToLoad() && $wgUser->isLoggedIn() && !$req ) {
 166                         $req = $this->getUserVariant();
 167                 } elseif ( !$req ) {
 168                         $req = $this->getHeaderVariant();
 169                 }
 170
 171                 if ( $wgDefaultLanguageVariant && !$req ) {
 172                         $req = $this->validateVariant( $wgDefaultLanguageVariant );
 173                 }
 174
 175                 // This function, unlike the other get*Variant functions, is
 176                 // not memoized (i.e. there return value is not cached) since
 177                 // new information might appear during processing after this
 178                 // is first called.
 179                 if ( $this->validateVariant( $req ) ) {
 180                         return $req;
 181                 }
 182                 return $this->mMainLanguageCode;
 183         }
 184
 185         /**
 186          * Get default variant.
 187          * This function would not be affected by user's settings
 188          * @return string The default variant code
 189          */
 190         public function getDefaultVariant() {
 191                 global $wgDefaultLanguageVariant;
 192
 193                 $req = $this->getURLVariant();
 194
 195                 if ( !$req ) {
 196                         $req = $this->getHeaderVariant();
 197                 }
 198
 199                 if ( $wgDefaultLanguageVariant && !$req ) {
 200                         $req = $this->validateVariant( $wgDefaultLanguageVariant );
 201                 }
 202
 203                 if ( $req ) {
 204                         return $req;
 205                 }
 206                 return $this->mMainLanguageCode;
 207         }
 208
 209         /**
 210          * Validate the variant
 211          * @param string $variant The variant to validate
 212          * @return mixed Returns the variant if it is valid, null otherwise
 213          */
 214         public function validateVariant( $variant = null ) {
 215                 if ( $variant !== null && in_array( $variant, $this->mVariants ) ) {
 216                         return $variant;
 217                 }
 218                 return null;
 219         }
 220
 221         /**
 222          * Get the variant specified in the URL
 223          *
 224          * @return mixed Variant if one found, false otherwise.
 225          */
 226         public function getURLVariant() {
 227                 global $wgRequest;
 228
 229                 if ( $this->mURLVariant ) {
 230                         return $this->mURLVariant;
 231                 }
 232
 233                 // see if the preference is set in the request
 234                 $ret = $wgRequest->getText( 'variant' );
 235
 236                 if ( !$ret ) {
 237                         $ret = $wgRequest->getVal( 'uselang' );
 238                 }
 239
 240                 $this->mURLVariant = $this->validateVariant( $ret );
 241                 return $this->mURLVariant;
 242         }
 243
 244         /**
 245          * Determine if the user has a variant set.
 246          *
 247          * @return mixed Variant if one found, false otherwise.
 248          */
 249         protected function getUserVariant() {
 250                 global $wgUser, $wgContLang;
 251
 252                 // memoizing this function wreaks havoc on parserTest.php
 253                 /*
 254                 if ( $this->mUserVariant ) {
 255                         return $this->mUserVariant;
 256                 }
 257                 */
 258
 259                 // Get language variant preference from logged in users
 260                 // Don't call this on stub objects because that causes infinite
 261                 // recursion during initialisation
 262                 if ( !$wgUser->isSafeToLoad() ) {
 263                         return false;
 264                 }
 265                 if ( $wgUser->isLoggedIn() ) {
 266                         if ( $this->mMainLanguageCode == $wgContLang->getCode() ) {
 267                                 $ret = $wgUser->getOption( 'variant' );
 268                         } else {
 269                                 $ret = $wgUser->getOption( 'variant-' . $this->mMainLanguageCode );
 270                         }
 271                 } else {
 272                         // figure out user lang without constructing wgLang to avoid
 273                         // infinite recursion
 274                         $ret = $wgUser->getOption( 'language' );
 275                 }
 276
 277                 $this->mUserVariant = $this->validateVariant( $ret );
 278                 return $this->mUserVariant;
 279         }
 280
 281         /**
 282          * Determine the language variant from the Accept-Language header.
 283          *
 284          * @return mixed Variant if one found, false otherwise.
 285          */
 286         protected function getHeaderVariant() {
 287                 global $wgRequest;
 288
 289                 if ( $this->mHeaderVariant ) {
 290                         return $this->mHeaderVariant;
 291                 }
 292
 293                 // see if some supported language variant is set in the
 294                 // HTTP header.
 295                 $languages = array_keys( $wgRequest->getAcceptLang() );
 296                 if ( empty( $languages ) ) {
 297                         return null;
 298                 }
 299
 300                 $fallbackLanguages = [];
 301                 foreach ( $languages as $language ) {
 302                         $this->mHeaderVariant = $this->validateVariant( $language );
 303                         if ( $this->mHeaderVariant ) {
 304                                 break;
 305                         }
 306
 307                         // To see if there are fallbacks of current language.
 308                         // We record these fallback variants, and process
 309                         // them later.
 310                         $fallbacks = $this->getVariantFallbacks( $language );
 311                         if ( is_string( $fallbacks ) && $fallbacks !== $this->mMainLanguageCode ) {
 312                                 $fallbackLanguages[] = $fallbacks;
 313                         } elseif ( is_array( $fallbacks ) ) {
 314                                 $fallbackLanguages =
 315                                         array_merge( $fallbackLanguages, $fallbacks );
 316                         }
 317                 }
 318
 319                 if ( !$this->mHeaderVariant ) {
 320                         // process fallback languages now
 321                         $fallback_languages = array_unique( $fallbackLanguages );
 322                         foreach ( $fallback_languages as $language ) {
 323                                 $this->mHeaderVariant = $this->validateVariant( $language );
 324                                 if ( $this->mHeaderVariant ) {
 325                                         break;
 326                                 }
 327                         }
 328                 }
 329
 330                 return $this->mHeaderVariant;
 331         }
 332
 333         /**
 334          * Dictionary-based conversion.
 335          * This function would not parse the conversion rules.
 336          * If you want to parse rules, try to use convert() or
 337          * convertTo().
 338          *
 339          * @param string $text The text to be converted
 340          * @param bool|string $toVariant The target language code
 341          * @return string The converted text
 342          */
 343         public function autoConvert( $text, $toVariant = false ) {
 344                 $this->loadTables();
 345
 346                 if ( !$toVariant ) {
 347                         $toVariant = $this->getPreferredVariant();
 348                         if ( !$toVariant ) {
 349                                 return $text;
 350                         }
 351                 }
 352
 353                 if ( $this->guessVariant( $text, $toVariant ) ) {
 354                         return $text;
 355                 }
 356
 357                 /* we convert everything except:
 358                  * 1. HTML markups (anything between < and >)
 359                  * 2. HTML entities
 360                  * 3. placeholders created by the parser
 361                  */
 362                 $marker = '|' . Parser::MARKER_PREFIX . '[\-a-zA-Z0-9]+';
 363
 364                 // this one is needed when the text is inside an HTML markup
 365                 $htmlfix = '|<[^>]+$|^[^<>]*>';
 366
 367                 // disable convert to variants between <code> tags
 368                 $codefix = '<code>.+?<\/code>|';
 369                 // disable conversion of <script> tags
 370                 $scriptfix = '<script.*?>.*?<\/script>|';
 371                 // disable conversion of <pre> tags
 372                 $prefix = '<pre.*?>.*?<\/pre>|';
 373
 374                 $reg = '/' . $codefix . $scriptfix . $prefix .
 375                         '<[^>]+>|&[a-zA-Z#][a-z0-9]+;' . $marker . $htmlfix . '/s';
 376                 $startPos = 0;
 377                 $sourceBlob = '';
 378                 $literalBlob = '';
 379
 380                 // Guard against delimiter nulls in the input
 381                 // (should never happen: see T159174)
 382                 $text = str_replace( "\000", '', $text );
 383
 384                 $markupMatches = null;
 385                 $elementMatches = null;
 386                 while ( $startPos < strlen( $text ) ) {
 387                         if ( preg_match( $reg, $text, $markupMatches, PREG_OFFSET_CAPTURE, $startPos ) ) {
 388                                 $elementPos = $markupMatches[0][1];
 389                                 $element = $markupMatches[0][0];
 390                         } else {
 391                                 $elementPos = strlen( $text );
 392                                 $element = '';
 393                         }
 394
 395                         // Queue the part before the markup for translation in a batch
 396                         $sourceBlob .= substr( $text, $startPos, $elementPos - $startPos ) . "\000";
 397
 398                         // Advance to the next position
 399                         $startPos = $elementPos + strlen( $element );
 400
 401                         // Translate any alt or title attributes inside the matched element
 402                         if ( $element !== ''
 403                                 && preg_match( '/^(<[^>\s]*)\s([^>]*)(.*)$/', $element, $elementMatches )
 404                         ) {
 405                                 $attrs = Sanitizer::decodeTagAttributes( $elementMatches[2] );
 406                                 $changed = false;
 407                                 foreach ( [ 'title', 'alt' ] as $attrName ) {
 408                                         if ( !isset( $attrs[$attrName] ) ) {
 409                                                 continue;
 410                                         }
 411                                         $attr = $attrs[$attrName];
 412                                         // Don't convert URLs
 413                                         if ( !strpos( $attr, '://' ) ) {
 414                                                 $attr = $this->recursiveConvertTopLevel( $attr, $toVariant );
 415                                         }
 416
 417                                         if ( $attr !== $attrs[$attrName] ) {
 418                                                 $attrs[$attrName] = $attr;
 419                                                 $changed = true;
 420                                         }
 421                                 }
 422                                 if ( $changed ) {
 423                                         $element = $elementMatches[1] . Html::expandAttributes( $attrs ) .
 424                                                 $elementMatches[3];
 425                                 }
 426                         }
 427                         $literalBlob .= $element . "\000";
 428                 }
 429
 430                 // Do the main translation batch
 431                 $translatedBlob = $this->translate( $sourceBlob, $toVariant );
 432
 433                 // Put the output back together
 434                 $translatedIter = StringUtils::explode( "\000", $translatedBlob );
 435                 $literalIter = StringUtils::explode( "\000", $literalBlob );
 436                 $output = '';
 437                 while ( $translatedIter->valid() && $literalIter->valid() ) {
 438                         $output .= $translatedIter->current();
 439                         $output .= $literalIter->current();
 440                         $translatedIter->next();
 441                         $literalIter->next();
 442                 }
 443
 444                 return $output;
 445         }
 446
 447         /**
 448          * Translate a string to a variant.
 449          * Doesn't parse rules or do any of that other stuff, for that use
 450          * convert() or convertTo().
 451          *
 452          * @param string $text Text to convert
 453          * @param string $variant Variant language code
 454          * @return string Translated text
 455          */
 456         public function translate( $text, $variant ) {
 457                 // If $text is empty or only includes spaces, do nothing
 458                 // Otherwise translate it
 459                 if ( trim( $text ) ) {
 460                         $this->loadTables();
 461                         $text = $this->mTables[$variant]->replace( $text );
 462                 }
 463                 return $text;
 464         }
 465
 466         /**
 467          * Call translate() to convert text to all valid variants.
 468          *
 469          * @param string $text The text to be converted
 470          * @return array Variant => converted text
 471          */
 472         public function autoConvertToAllVariants( $text ) {
 473                 $this->loadTables();
 474
 475                 $ret = [];
 476                 foreach ( $this->mVariants as $variant ) {
 477                         $ret[$variant] = $this->translate( $text, $variant );
 478                 }
 479
 480                 return $ret;
 481         }
 482
 483         /**
 484          * Apply manual conversion rules.
 485          *
 486          * @param ConverterRule $convRule
 487          */
 488         protected function applyManualConv( $convRule ) {
 489                 // Use syntax -{T|zh-cn:TitleCN; zh-tw:TitleTw}- to custom
 490                 // title conversion.
 491                 // T26072: $mConvRuleTitle was overwritten by other manual
 492                 // rule(s) not for title, this breaks the title conversion.
 493                 $newConvRuleTitle = $convRule->getTitle();
 494                 if ( $newConvRuleTitle ) {
 495                         // So I add an empty check for getTitle()
 496                         $this->mConvRuleTitle = $newConvRuleTitle;
 497                 }
 498
 499                 // merge/remove manual conversion rules to/from global table
 500                 $convTable = $convRule->getConvTable();
 501                 $action = $convRule->getRulesAction();
 502                 foreach ( $convTable as $variant => $pair ) {
 503                         if ( !$this->validateVariant( $variant ) ) {
 504                                 continue;
 505                         }
 506
 507                         if ( $action == 'add' ) {
 508                                 // More efficient than array_merge(), about 2.5 times.
 509                                 foreach ( $pair as $from => $to ) {
 510                                         $this->mTables[$variant]->setPair( $from, $to );
 511                                 }
 512                         } elseif ( $action == 'remove' ) {
 513                                 $this->mTables[$variant]->removeArray( $pair );
 514                         }
 515                 }
 516         }
 517
 518         /**
 519          * Auto convert a Title object to a readable string in the
 520          * preferred variant.
 521          *
 522          * @param Title $title A object of Title
 523          * @return string Converted title text
 524          */
 525         public function convertTitle( $title ) {
 526                 $variant = $this->getPreferredVariant();
 527                 $index = $title->getNamespace();
 528                 if ( $index !== NS_MAIN ) {
 529                         $text = $this->convertNamespace( $index, $variant ) . ':';
 530                 } else {
 531                         $text = '';
 532                 }
 533                 $text .= $this->translate( $title->getText(), $variant );
 534                 return $text;
 535         }
 536
 537         /**
 538          * Get the namespace display name in the preferred variant.
 539          *
 540          * @param int $index Namespace id
 541          * @param string|null $variant Variant code or null for preferred variant
 542          * @return string Namespace name for display
 543          */
 544         public function convertNamespace( $index, $variant = null ) {
 545                 if ( $index === NS_MAIN ) {
 546                         return '';
 547                 }
 548
 549                 if ( $variant === null ) {
 550                         $variant = $this->getPreferredVariant();
 551                 }
 552
 553                 $cache = MediaWikiServices::getInstance()->getLocalServerObjectCache();
 554                 $key = $cache->makeKey( 'languageconverter', 'namespace-text', $index, $variant );
 555                 $nsVariantText = $cache->get( $key );
 556                 if ( $nsVariantText !== false ) {
 557                         return $nsVariantText;
 558                 }
 559
 560                 // First check if a message gives a converted name in the target variant.
 561                 $nsConvMsg = wfMessage( 'conversion-ns' . $index )->inLanguage( $variant );
 562                 if ( $nsConvMsg->exists() ) {
 563                         $nsVariantText = $nsConvMsg->plain();
 564                 }
 565
 566                 // Then check if a message gives a converted name in content language
 567                 // which needs extra translation to the target variant.
 568                 if ( $nsVariantText === false ) {
 569                         $nsConvMsg = wfMessage( 'conversion-ns' . $index )->inContentLanguage();
 570                         if ( $nsConvMsg->exists() ) {
 571                                 $nsVariantText = $this->translate( $nsConvMsg->plain(), $variant );
 572                         }
 573                 }
 574
 575                 if ( $nsVariantText === false ) {
 576                         // No message exists, retrieve it from the target variant's namespace names.
 577                         $langObj = $this->mLangObj->factory( $variant );
 578                         $nsVariantText = $langObj->getFormattedNsText( $index );
 579                 }
 580
 581                 $cache->set( $key, $nsVariantText, 60 );
 582
 583                 return $nsVariantText;
 584         }
 585
 586         /**
 587          * Convert text to different variants of a language. The automatic
 588          * conversion is done in autoConvert(). Here we parse the text
 589          * marked with -{}-, which specifies special conversions of the
 590          * text that can not be accomplished in autoConvert().
 591          *
 592          * Syntax of the markup:
 593          * -{code1:text1;code2:text2;...}-  or
 594          * -{flags|code1:text1;code2:text2;...}-  or
 595          * -{text}- in which case no conversion should take place for text
 596          *
 597          * @param string $text Text to be converted
 598          * @return string Converted text
 599          */
 600         public function convert( $text ) {
 601                 $variant = $this->getPreferredVariant();
 602                 return $this->convertTo( $text, $variant );
 603         }
 604
 605         /**
 606          * Same as convert() except a extra parameter to custom variant.
 607          *
 608          * @param string $text Text to be converted
 609          * @param string $variant The target variant code
 610          * @return string Converted text
 611          */
 612         public function convertTo( $text, $variant ) {
 613                 global $wgDisableLangConversion;
 614                 if ( $wgDisableLangConversion ) {
 615                         return $text;
 616                 }
 617                 // Reset converter state for a new converter run.
 618                 $this->mConvRuleTitle = false;
 619                 return $this->recursiveConvertTopLevel( $text, $variant );
 620         }
 621
 622         /**
 623          * Recursively convert text on the outside. Allow to use nested
 624          * markups to custom rules.
 625          *
 626          * @param string $text Text to be converted
 627          * @param string $variant The target variant code
 628          * @param int $depth Depth of recursion
 629          * @return string Converted text
 630          */
 631         protected function recursiveConvertTopLevel( $text, $variant, $depth = 0 ) {
 632                 $startPos = 0;
 633                 $out = '';
 634                 $length = strlen( $text );
 635                 $shouldConvert = !$this->guessVariant( $text, $variant );
 636
 637                 while ( $startPos < $length ) {
 638                         $pos = strpos( $text, '-{', $startPos );
 639
 640                         if ( $pos === false ) {
 641                                 // No more markup, append final segment
 642                                 $fragment = substr( $text, $startPos );
 643                                 $out .= $shouldConvert ? $this->autoConvert( $fragment, $variant ) : $fragment;
 644                                 return $out;
 645                         }
 646
 647                         // Markup found
 648                         // Append initial segment
 649                         $fragment = substr( $text, $startPos, $pos - $startPos );
 650                         $out .= $shouldConvert ? $this->autoConvert( $fragment, $variant ) : $fragment;
 651
 652                         // Advance position
 653                         $startPos = $pos;
 654
 655                         // Do recursive conversion
 656                         $out .= $this->recursiveConvertRule( $text, $variant, $startPos, $depth + 1 );
 657                 }
 658
 659                 return $out;
 660         }
 661
 662         /**
 663          * Recursively convert text on the inside.
 664          *
 665          * @param string $text Text to be converted
 666          * @param string $variant The target variant code
 667          * @param int $startPos
 668          * @param int $depth Depth of recursion
 669          *
 670          * @throws MWException
 671          * @return string Converted text
 672          */
 673         protected function recursiveConvertRule( $text, $variant, &$startPos, $depth = 0 ) {
 674                 // Quick sanity check (no function calls)
 675                 if ( $text[$startPos] !== '-' || $text[$startPos + 1] !== '{' ) {
 676                         throw new MWException( __METHOD__ . ': invalid input string' );
 677                 }
 678
 679                 $startPos += 2;
 680                 $inner = '';
 681                 $warningDone = false;
 682                 $length = strlen( $text );
 683
 684                 while ( $startPos < $length ) {
 685                         $m = false;
 686                         preg_match( '/-\{|\}-/', $text, $m, PREG_OFFSET_CAPTURE, $startPos );
 687                         if ( !$m ) {
 688                                 // Unclosed rule
 689                                 break;
 690                         }
 691
 692                         $token = $m[0][0];
 693                         $pos = $m[0][1];
 694
 695                         // Markup found
 696                         // Append initial segment
 697                         $inner .= substr( $text, $startPos, $pos - $startPos );
 698
 699                         // Advance position
 700                         $startPos = $pos;
 701
 702                         switch ( $token ) {
 703                                 case '-{':
 704                                         // Check max depth
 705                                         if ( $depth >= $this->mMaxDepth ) {
 706                                                 $inner .= '-{';
 707                                                 if ( !$warningDone ) {
 708                                                         $inner .= '<span class="error">' .
 709                                                                 wfMessage( 'language-converter-depth-warning' )
 710                                                                         ->numParams( $this->mMaxDepth )->inContentLanguage()->text() .
 711                                                                 '</span>';
 712                                                         $warningDone = true;
 713                                                 }
 714                                                 $startPos += 2;
 715                                                 continue;
 716                                         }
 717                                         // Recursively parse another rule
 718                                         $inner .= $this->recursiveConvertRule( $text, $variant, $startPos, $depth + 1 );
 719                                         break;
 720                                 case '}-':
 721                                         // Apply the rule
 722                                         $startPos += 2;
 723                                         $rule = new ConverterRule( $inner, $this );
 724                                         $rule->parse( $variant );
 725                                         $this->applyManualConv( $rule );
 726                                         return $rule->getDisplay();
 727                                 default:
 728                                         throw new MWException( __METHOD__ . ': invalid regex match' );
 729                         }
 730                 }
 731
 732                 // Unclosed rule
 733                 if ( $startPos < $length ) {
 734                         $inner .= substr( $text, $startPos );
 735                 }
 736                 $startPos = $length;
 737                 return '-{' . $this->autoConvert( $inner, $variant );
 738         }
 739
 740         /**
 741          * If a language supports multiple variants, it is possible that
 742          * non-existing link in one variant actually exists in another variant.
 743          * This function tries to find it. See e.g. LanguageZh.php
 744          * The input parameters may be modified upon return
 745          *
 746          * @param string &$link The name of the link
 747          * @param Title &$nt The title object of the link
 748          * @param bool $ignoreOtherCond To disable other conditions when
 749          *   we need to transclude a template or update a category's link
 750          */
 751         public function findVariantLink( &$link, &$nt, $ignoreOtherCond = false ) {
 752                 # If the article has already existed, there is no need to
 753                 # check it again, otherwise it may cause a fault.
 754                 if ( is_object( $nt ) && $nt->exists() ) {
 755                         return;
 756                 }
 757
 758                 global $wgDisableLangConversion, $wgDisableTitleConversion, $wgRequest;
 759                 $isredir = $wgRequest->getText( 'redirect', 'yes' );
 760                 $action = $wgRequest->getText( 'action' );
 761                 if ( $action == 'edit' && $wgRequest->getBool( 'redlink' ) ) {
 762                         $action = 'view';
 763                 }
 764                 $linkconvert = $wgRequest->getText( 'linkconvert', 'yes' );
 765                 $disableLinkConversion = $wgDisableLangConversion
 766                         || $wgDisableTitleConversion;
 767                 $linkBatch = new LinkBatch();
 768
 769                 $ns = NS_MAIN;
 770
 771                 if ( $disableLinkConversion ||
 772                         ( !$ignoreOtherCond &&
 773                                 ( $isredir == 'no'
 774                                         || $action == 'edit'
 775                                         || $action == 'submit'
 776                                         || $linkconvert == 'no' ) ) ) {
 777                         return;
 778                 }
 779
 780                 if ( is_object( $nt ) ) {
 781                         $ns = $nt->getNamespace();
 782                 }
 783
 784                 $variants = $this->autoConvertToAllVariants( $link );
 785                 if ( !$variants ) { // give up
 786                         return;
 787                 }
 788
 789                 $titles = [];
 790
 791                 foreach ( $variants as $v ) {
 792                         if ( $v != $link ) {
 793                                 $varnt = Title::newFromText( $v, $ns );
 794                                 if ( !is_null( $varnt ) ) {
 795                                         $linkBatch->addObj( $varnt );
 796                                         $titles[] = $varnt;
 797                                 }
 798                         }
 799                 }
 800
 801                 // fetch all variants in single query
 802                 $linkBatch->execute();
 803
 804                 foreach ( $titles as $varnt ) {
 805                         if ( $varnt->getArticleID() > 0 ) {
 806                                 $nt = $varnt;
 807                                 $link = $varnt->getText();
 808                                 break;
 809                         }
 810                 }
 811         }
 812
 813         /**
 814          * Returns language specific hash options.
 815          *
 816          * @return string
 817          */
 818         public function getExtraHashOptions() {
 819                 $variant = $this->getPreferredVariant();
 820
 821                 return '!' . $variant;
 822         }
 823
 824         /**
 825          * Guess if a text is written in a variant. This should be implemented in subclasses.
 826          *
 827          * @param string $text The text to be checked
 828          * @param string $variant Language code of the variant to be checked for
 829          * @return bool True if $text appears to be written in $variant, false if not
 830          *
 831          * @author Nikola Smolenski <smolensk@eunet.rs>
 832          * @since 1.19
 833          */
 834         public function guessVariant( $text, $variant ) {
 835                 return false;
 836         }
 837
 838         /**
 839          * Load default conversion tables.
 840          * This method must be implemented in derived class.
 841          *
 842          * @private
 843          * @throws MWException
 844          */
 845         function loadDefaultTables() {
 846                 $class = static::class;
 847                 throw new MWException( "Must implement loadDefaultTables() method in class $class" );
 848         }
 849
 850         /**
 851          * Load conversion tables either from the cache or the disk.
 852          * @private
 853          * @param bool $fromCache Load from memcached? Defaults to true.
 854          */
 855         function loadTables( $fromCache = true ) {
 856                 global $wgLanguageConverterCacheType;
 857
 858                 if ( $this->mTablesLoaded ) {
 859                         return;
 860                 }
 861
 862                 $this->mTablesLoaded = true;
 863                 $this->mTables = false;
 864                 $cache = ObjectCache::getInstance( $wgLanguageConverterCacheType );
 865                 $cacheKey = $cache->makeKey( 'conversiontables', $this->mMainLanguageCode );
 866                 if ( $fromCache ) {
 867                         $this->mTables = $cache->get( $cacheKey );
 868                 }
 869                 if ( !$this->mTables || !array_key_exists( self::CACHE_VERSION_KEY, $this->mTables ) ) {
 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( $cacheKey, $this->mTables, 43200 );
 883                 }
 884         }
 885
 886         /**
 887          * Hook for post processing after conversion tables are loaded.
 888          */
 889         function postLoadTables() {
 890         }
 891
 892         /**
 893          * Reload the conversion tables.
 894          *
 895          * Also used by test suites which need to reset the converter state.
 896          *
 897          * @private
 898          */
 899         private function reloadTables() {
 900                 if ( $this->mTables ) {
 901                         unset( $this->mTables );
 902                 }
 903
 904                 $this->mTablesLoaded = false;
 905                 $this->loadTables( false );
 906         }
 907
 908         /**
 909          * Parse the conversion table stored in the cache.
 910          *
 911          * The tables should be in blocks of the following form:
 912          *              -{
 913          *                      word => word ;
 914          *                      word => word ;
 915          *                      ...
 916          *              }-
 917          *
 918          * To make the tables more manageable, subpages are allowed
 919          * and will be parsed recursively if $recursive == true.
 920          *
 921          * @param string $code Language code
 922          * @param string $subpage Subpage name
 923          * @param bool $recursive Parse subpages recursively? Defaults to true.
 924          *
 925          * @return array
 926          */
 927         function parseCachedTable( $code, $subpage = '', $recursive = true ) {
 928                 static $parsed = [];
 929
 930                 $key = 'Conversiontable/' . $code;
 931                 if ( $subpage ) {
 932                         $key .= '/' . $subpage;
 933                 }
 934                 if ( array_key_exists( $key, $parsed ) ) {
 935                         return [];
 936                 }
 937
 938                 $parsed[$key] = true;
 939
 940                 if ( $subpage === '' ) {
 941                         $txt = MessageCache::singleton()->getMsgFromNamespace( $key, $code );
 942                 } else {
 943                         $txt = false;
 944                         $title = Title::makeTitleSafe( NS_MEDIAWIKI, $key );
 945                         if ( $title && $title->exists() ) {
 946                                 $revision = Revision::newFromTitle( $title );
 947                                 if ( $revision ) {
 948                                         if ( $revision->getContentModel() == CONTENT_MODEL_WIKITEXT ) {
 949                                                 $txt = $revision->getContent( Revision::RAW )->getNativeData();
 950                                         }
 951
 952                                         // @todo in the future, use a specialized content model, perhaps based on json!
 953                                 }
 954                         }
 955                 }
 956
 957                 # Nothing to parse if there's no text
 958                 if ( $txt === false || $txt === null || $txt === '' ) {
 959                         return [];
 960                 }
 961
 962                 // get all subpage links of the form
 963                 // [[MediaWiki:Conversiontable/zh-xx/...|...]]
 964                 $linkhead = $this->mLangObj->getNsText( NS_MEDIAWIKI ) .
 965                         ':Conversiontable';
 966                 $subs = StringUtils::explode( '[[', $txt );
 967                 $sublinks = [];
 968                 foreach ( $subs as $sub ) {
 969                         $link = explode( ']]', $sub, 2 );
 970                         if ( count( $link ) != 2 ) {
 971                                 continue;
 972                         }
 973                         $b = explode( '|', $link[0], 2 );
 974                         $b = explode( '/', trim( $b[0] ), 3 );
 975                         if ( count( $b ) == 3 ) {
 976                                 $sublink = $b[2];
 977                         } else {
 978                                 $sublink = '';
 979                         }
 980
 981                         if ( $b[0] == $linkhead && $b[1] == $code ) {
 982                                 $sublinks[] = $sublink;
 983                         }
 984                 }
 985
 986                 // parse the mappings in this page
 987                 $blocks = StringUtils::explode( '-{', $txt );
 988                 $ret = [];
 989                 $first = true;
 990                 foreach ( $blocks as $block ) {
 991                         if ( $first ) {
 992                                 // Skip the part before the first -{
 993                                 $first = false;
 994                                 continue;
 995                         }
 996                         $mappings = explode( '}-', $block, 2 )[0];
 997                         $stripped = str_replace( [ "'", '"', '*', '#' ], '', $mappings );
 998                         $table = StringUtils::explode( ';', $stripped );
 999                         foreach ( $table as $t ) {
1000                                 $m = explode( '=>', $t, 3 );
1001                                 if ( count( $m ) != 2 ) {
1002                                         continue;
1003                                 }
1004                                 // trim any trailling comments starting with '//'
1005                                 $tt = explode( '//', $m[1], 2 );
1006                                 $ret[trim( $m[0] )] = trim( $tt[0] );
1007                         }
1008                 }
1009
1010                 // recursively parse the subpages
1011                 if ( $recursive ) {
1012                         foreach ( $sublinks as $link ) {
1013                                 $s = $this->parseCachedTable( $code, $link, $recursive );
1014                                 $ret = $s + $ret;
1015                         }
1016                 }
1017
1018                 if ( $this->mUcfirst ) {
1019                         foreach ( $ret as $k => $v ) {
1020                                 $ret[$this->mLangObj->ucfirst( $k )] = $this->mLangObj->ucfirst( $v );
1021                         }
1022                 }
1023                 return $ret;
1024         }
1025
1026         /**
1027          * Enclose a string with the "no conversion" tag. This is used by
1028          * various functions in the Parser.
1029          *
1030          * @param string $text Text to be tagged for no conversion
1031          * @param bool $noParse Unused
1032          * @return string The tagged text
1033          */
1034         public function markNoConversion( $text, $noParse = false ) {
1035                 # don't mark if already marked
1036                 if ( strpos( $text, '-{' ) || strpos( $text, '}-' ) ) {
1037                         return $text;
1038                 }
1039
1040                 $ret = "-{R|$text}-";
1041                 return $ret;
1042         }
1043
1044         /**
1045          * Convert the sorting key for category links. This should make different
1046          * keys that are variants of each other map to the same key.
1047          *
1048          * @param string $key
1049          *
1050          * @return string
1051          */
1052         function convertCategoryKey( $key ) {
1053                 return $key;
1054         }
1055
1056         /**
1057          * Refresh the cache of conversion tables when
1058          * MediaWiki:Conversiontable* is updated.
1059          *
1060          * @param Title $titleobj The Title of the page being updated
1061          */
1062         public function updateConversionTable( Title $titleobj ) {
1063                 if ( $titleobj->getNamespace() == NS_MEDIAWIKI ) {
1064                         $title = $titleobj->getDBkey();
1065                         $t = explode( '/', $title, 3 );
1066                         $c = count( $t );
1067                         if ( $c > 1 && $t[0] == 'Conversiontable' ) {
1068                                 if ( $this->validateVariant( $t[1] ) ) {
1069                                         $this->reloadTables();
1070                                 }
1071                         }
1072                 }
1073         }
1074
1075         /**
1076          * Get the cached separator pattern for ConverterRule::parseRules()
1077          * @return string
1078          */
1079         function getVarSeparatorPattern() {
1080                 if ( is_null( $this->mVarSeparatorPattern ) ) {
1081                         // varsep_pattern for preg_split:
1082                         // text should be splited by ";" only if a valid variant
1083                         // name exist after the markup, for example:
1084                         //  -{zh-hans:<span style="font-size:120%;">xxx</span>;zh-hant:\
1085                         //  <span style="font-size:120%;">yyy</span>;}-
1086                         // we should split it as:
1087                         //  [
1088                         //    [0] => 'zh-hans:<span style="font-size:120%;">xxx</span>'
1089                         //    [1] => 'zh-hant:<span style="font-size:120%;">yyy</span>'
1090                         //    [2] => ''
1091                         //  ]
1092                         $pat = '/;\s*(?=';
1093                         foreach ( $this->mVariants as $variant ) {
1094                                 // zh-hans:xxx;zh-hant:yyy
1095                                 $pat .= $variant . '\s*:|';
1096                                 // xxx=>zh-hans:yyy; xxx=>zh-hant:zzz
1097                                 $pat .= '[^;]*?=>\s*' . $variant . '\s*:|';
1098                         }
1099                         $pat .= '\s*$)/';
1100                         $this->mVarSeparatorPattern = $pat;
1101                 }
1102                 return $this->mVarSeparatorPattern;
1103         }
1104 }