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                 'gan',
  40                 'iu',
  41                 'kk',
  42                 'ku',
  43                 'shi',
  44                 'sr',
  45                 'tg',
  46                 'uz',
  47                 'zh',
  48         ];
  49
  50         public $mMainLanguageCode;
  51
  52         /**
  53          * @var string[]
  54          */
  55         public $mVariants;
  56         public $mVariantFallbacks;
  57         public $mVariantNames;
  58         public $mTablesLoaded = false;
  59         public $mTables;
  60         // 'bidirectional' 'unidirectional' 'disable' for each variant
  61         public $mManualLevel;
  62
  63         /**
  64          * @var string Memcached key name
  65          */
  66         public $mCacheKey;
  67
  68         public $mLangObj;
  69         public $mFlags;
  70         public $mDescCodeSep = ':', $mDescVarSep = ';';
  71         public $mUcfirst = false;
  72         public $mConvRuleTitle = false;
  73         public $mURLVariant;
  74         public $mUserVariant;
  75         public $mHeaderVariant;
  76         public $mMaxDepth = 10;
  77         public $mVarSeparatorPattern;
  78
  79         const CACHE_VERSION_KEY = 'VERSION 7';
  80
  81         /**
  82          * @param Language $langobj
  83          * @param string $maincode The main language code of this language
  84          * @param string[] $variants The supported variants of this language
  85          * @param array $variantfallbacks The fallback language of each variant
  86          * @param array $flags Defining the custom strings that maps to the flags
  87          * @param array $manualLevel Limit for supported variants
  88          */
  89         public function __construct( $langobj, $maincode, $variants = [],
  90                                                                 $variantfallbacks = [], $flags = [],
  91                                                                 $manualLevel = [] ) {
  92                 global $wgDisabledVariants;
  93                 $this->mLangObj = $langobj;
  94                 $this->mMainLanguageCode = $maincode;
  95                 $this->mVariants = array_diff( $variants, $wgDisabledVariants );
  96                 $this->mVariantFallbacks = $variantfallbacks;
  97                 $this->mVariantNames = Language::fetchLanguageNames();
  98                 $this->mCacheKey = wfMemcKey( 'conversiontables', $maincode );
  99                 $defaultflags = [
 100                         // 'S' show converted text
 101                         // '+' add rules for alltext
 102                         // 'E' the gave flags is error
 103                         // these flags above are reserved for program
 104                         'A' => 'A',   // add rule for convert code (all text convert)
 105                         'T' => 'T',   // title convert
 106                         'R' => 'R',   // raw content
 107                         'D' => 'D',   // convert description (subclass implement)
 108                         '-' => '-',   // remove convert (not implement)
 109                         'H' => 'H',   // add rule for convert code (but no display in placed code)
 110                         'N' => 'N',   // current variant name
 111                 ];
 112                 $this->mFlags = array_merge( $defaultflags, $flags );
 113                 foreach ( $this->mVariants as $v ) {
 114                         if ( array_key_exists( $v, $manualLevel ) ) {
 115                                 $this->mManualLevel[$v] = $manualLevel[$v];
 116                         } else {
 117                                 $this->mManualLevel[$v] = 'bidirectional';
 118                         }
 119                         $this->mFlags[$v] = $v;
 120                 }
 121         }
 122
 123         /**
 124          * Get all valid variants.
 125          * Call this instead of using $this->mVariants directly.
 126          *
 127          * @return string[] Contains all valid variants
 128          */
 129         public function getVariants() {
 130                 return $this->mVariants;
 131         }
 132
 133         /**
 134          * In case some variant is not defined in the markup, we need
 135          * to have some fallback. For example, in zh, normally people
 136          * will define zh-hans and zh-hant, but less so for zh-sg or zh-hk.
 137          * when zh-sg is preferred but not defined, we will pick zh-hans
 138          * in this case. Right now this is only used by zh.
 139          *
 140          * @param string $variant The language code of the variant
 141          * @return string|array The code of the fallback language or the
 142          *   main code if there is no fallback
 143          */
 144         public function getVariantFallbacks( $variant ) {
 145                 if ( isset( $this->mVariantFallbacks[$variant] ) ) {
 146                         return $this->mVariantFallbacks[$variant];
 147                 }
 148                 return $this->mMainLanguageCode;
 149         }
 150
 151         /**
 152          * Get the title produced by the conversion rule.
 153          * @return string The converted title text
 154          */
 155         public function getConvRuleTitle() {
 156                 return $this->mConvRuleTitle;
 157         }
 158
 159         /**
 160          * Get preferred language variant.
 161          * @return string The preferred language code
 162          */
 163         public function getPreferredVariant() {
 164                 global $wgDefaultLanguageVariant, $wgUser;
 165
 166                 $req = $this->getURLVariant();
 167
 168                 if ( $wgUser->isSafeToLoad() && $wgUser->isLoggedIn() && !$req ) {
 169                         $req = $this->getUserVariant();
 170                 } elseif ( !$req ) {
 171                         $req = $this->getHeaderVariant();
 172                 }
 173
 174                 if ( $wgDefaultLanguageVariant && !$req ) {
 175                         $req = $this->validateVariant( $wgDefaultLanguageVariant );
 176                 }
 177
 178                 // This function, unlike the other get*Variant functions, is
 179                 // not memoized (i.e. there return value is not cached) since
 180                 // new information might appear during processing after this
 181                 // is first called.
 182                 if ( $this->validateVariant( $req ) ) {
 183                         return $req;
 184                 }
 185                 return $this->mMainLanguageCode;
 186         }
 187
 188         /**
 189          * Get default variant.
 190          * This function would not be affected by user's settings
 191          * @return string The default variant code
 192          */
 193         public function getDefaultVariant() {
 194                 global $wgDefaultLanguageVariant;
 195
 196                 $req = $this->getURLVariant();
 197
 198                 if ( !$req ) {
 199                         $req = $this->getHeaderVariant();
 200                 }
 201
 202                 if ( $wgDefaultLanguageVariant && !$req ) {
 203                         $req = $this->validateVariant( $wgDefaultLanguageVariant );
 204                 }
 205
 206                 if ( $req ) {
 207                         return $req;
 208                 }
 209                 return $this->mMainLanguageCode;
 210         }
 211
 212         /**
 213          * Validate the variant
 214          * @param string $variant The variant to validate
 215          * @return mixed Returns the variant if it is valid, null otherwise
 216          */
 217         public function validateVariant( $variant = null ) {
 218                 if ( $variant !== null && in_array( $variant, $this->mVariants ) ) {
 219                         return $variant;
 220                 }
 221                 return null;
 222         }
 223
 224         /**
 225          * Get the variant specified in the URL
 226          *
 227          * @return mixed Variant if one found, false otherwise.
 228          */
 229         public function getURLVariant() {
 230                 global $wgRequest;
 231
 232                 if ( $this->mURLVariant ) {
 233                         return $this->mURLVariant;
 234                 }
 235
 236                 // see if the preference is set in the request
 237                 $ret = $wgRequest->getText( 'variant' );
 238
 239                 if ( !$ret ) {
 240                         $ret = $wgRequest->getVal( 'uselang' );
 241                 }
 242
 243                 $this->mURLVariant = $this->validateVariant( $ret );
 244                 return $this->mURLVariant;
 245         }
 246
 247         /**
 248          * Determine if the user has a variant set.
 249          *
 250          * @return mixed Variant if one found, false otherwise.
 251          */
 252         protected function getUserVariant() {
 253                 global $wgUser, $wgContLang;
 254
 255                 // memoizing this function wreaks havoc on parserTest.php
 256                 /*
 257                 if ( $this->mUserVariant ) {
 258                         return $this->mUserVariant;
 259                 }
 260                 */
 261
 262                 // Get language variant preference from logged in users
 263                 // Don't call this on stub objects because that causes infinite
 264                 // recursion during initialisation
 265                 if ( !$wgUser->isSafeToLoad() ) {
 266                         return false;
 267                 }
 268                 if ( $wgUser->isLoggedIn() ) {
 269                         if ( $this->mMainLanguageCode == $wgContLang->getCode() ) {
 270                                 $ret = $wgUser->getOption( 'variant' );
 271                         } else {
 272                                 $ret = $wgUser->getOption( 'variant-' . $this->mMainLanguageCode );
 273                         }
 274                 } else {
 275                         // figure out user lang without constructing wgLang to avoid
 276                         // infinite recursion
 277                         $ret = $wgUser->getOption( 'language' );
 278                 }
 279
 280                 $this->mUserVariant = $this->validateVariant( $ret );
 281                 return $this->mUserVariant;
 282         }
 283
 284         /**
 285          * Determine the language variant from the Accept-Language header.
 286          *
 287          * @return mixed Variant if one found, false otherwise.
 288          */
 289         protected function getHeaderVariant() {
 290                 global $wgRequest;
 291
 292                 if ( $this->mHeaderVariant ) {
 293                         return $this->mHeaderVariant;
 294                 }
 295
 296                 // see if some supported language variant is set in the
 297                 // HTTP header.
 298                 $languages = array_keys( $wgRequest->getAcceptLang() );
 299                 if ( empty( $languages ) ) {
 300                         return null;
 301                 }
 302
 303                 $fallbackLanguages = [];
 304                 foreach ( $languages as $language ) {
 305                         $this->mHeaderVariant = $this->validateVariant( $language );
 306                         if ( $this->mHeaderVariant ) {
 307                                 break;
 308                         }
 309
 310                         // To see if there are fallbacks of current language.
 311                         // We record these fallback variants, and process
 312                         // them later.
 313                         $fallbacks = $this->getVariantFallbacks( $language );
 314                         if ( is_string( $fallbacks ) && $fallbacks !== $this->mMainLanguageCode ) {
 315                                 $fallbackLanguages[] = $fallbacks;
 316                         } elseif ( is_array( $fallbacks ) ) {
 317                                 $fallbackLanguages =
 318                                         array_merge( $fallbackLanguages, $fallbacks );
 319                         }
 320                 }
 321
 322                 if ( !$this->mHeaderVariant ) {
 323                         // process fallback languages now
 324                         $fallback_languages = array_unique( $fallbackLanguages );
 325                         foreach ( $fallback_languages as $language ) {
 326                                 $this->mHeaderVariant = $this->validateVariant( $language );
 327                                 if ( $this->mHeaderVariant ) {
 328                                         break;
 329                                 }
 330                         }
 331                 }
 332
 333                 return $this->mHeaderVariant;
 334         }
 335
 336         /**
 337          * Dictionary-based conversion.
 338          * This function would not parse the conversion rules.
 339          * If you want to parse rules, try to use convert() or
 340          * convertTo().
 341          *
 342          * @param string $text The text to be converted
 343          * @param bool|string $toVariant The target language code
 344          * @return string The converted text
 345          */
 346         public function autoConvert( $text, $toVariant = false ) {
 347
 348                 $this->loadTables();
 349
 350                 if ( !$toVariant ) {
 351                         $toVariant = $this->getPreferredVariant();
 352                         if ( !$toVariant ) {
 353                                 return $text;
 354                         }
 355                 }
 356
 357                 if ( $this->guessVariant( $text, $toVariant ) ) {
 358                         return $text;
 359                 }
 360
 361                 /* we convert everything except:
 362                  * 1. HTML markups (anything between < and >)
 363                  * 2. HTML entities
 364                  * 3. placeholders created by the parser
 365                  */
 366                 $marker = '|' . Parser::MARKER_PREFIX . '[\-a-zA-Z0-9]+';
 367
 368                 // this one is needed when the text is inside an HTML markup
 369                 $htmlfix = '|<[^>]+$|^[^<>]*>';
 370
 371                 // disable convert to variants between <code> tags
 372                 $codefix = '<code>.+?<\/code>|';
 373                 // disable conversion of <script> tags
 374                 $scriptfix = '<script.*?>.*?<\/script>|';
 375                 // disable conversion of <pre> tags
 376                 $prefix = '<pre.*?>.*?<\/pre>|';
 377
 378                 $reg = '/' . $codefix . $scriptfix . $prefix .
 379                         '<[^>]+>|&[a-zA-Z#][a-z0-9]+;' . $marker . $htmlfix . '/s';
 380                 $startPos = 0;
 381                 $sourceBlob = '';
 382                 $literalBlob = '';
 383
 384                 // Guard against delimiter nulls in the input
 385                 // (should never happen: see T159174)
 386                 $text = str_replace( "\000", '', $text );
 387
 388                 $markupMatches = null;
 389                 $elementMatches = null;
 390                 while ( $startPos < strlen( $text ) ) {
 391                         if ( preg_match( $reg, $text, $markupMatches, PREG_OFFSET_CAPTURE, $startPos ) ) {
 392                                 $elementPos = $markupMatches[0][1];
 393                                 $element = $markupMatches[0][0];
 394                         } else {
 395                                 $elementPos = strlen( $text );
 396                                 $element = '';
 397                         }
 398
 399                         // Queue the part before the markup for translation in a batch
 400                         $sourceBlob .= substr( $text, $startPos, $elementPos - $startPos ) . "\000";
 401
 402                         // Advance to the next position
 403                         $startPos = $elementPos + strlen( $element );
 404
 405                         // Translate any alt or title attributes inside the matched element
 406                         if ( $element !== ''
 407                                 && preg_match( '/^(<[^>\s]*)\s([^>]*)(.*)$/', $element, $elementMatches )
 408                         ) {
 409                                 $attrs = Sanitizer::decodeTagAttributes( $elementMatches[2] );
 410                                 $changed = false;
 411                                 foreach ( [ 'title', 'alt' ] as $attrName ) {
 412                                         if ( !isset( $attrs[$attrName] ) ) {
 413                                                 continue;
 414                                         }
 415                                         $attr = $attrs[$attrName];
 416                                         // Don't convert URLs
 417                                         if ( !strpos( $attr, '://' ) ) {
 418                                                 $attr = $this->recursiveConvertTopLevel( $attr, $toVariant );
 419                                         }
 420
 421                                         if ( $attr !== $attrs[$attrName] ) {
 422                                                 $attrs[$attrName] = $attr;
 423                                                 $changed = true;
 424                                         }
 425                                 }
 426                                 if ( $changed ) {
 427                                         $element = $elementMatches[1] . Html::expandAttributes( $attrs ) .
 428                                                 $elementMatches[3];
 429                                 }
 430                         }
 431                         $literalBlob .= $element . "\000";
 432                 }
 433
 434                 // Do the main translation batch
 435                 $translatedBlob = $this->translate( $sourceBlob, $toVariant );
 436
 437                 // Put the output back together
 438                 $translatedIter = StringUtils::explode( "\000", $translatedBlob );
 439                 $literalIter = StringUtils::explode( "\000", $literalBlob );
 440                 $output = '';
 441                 while ( $translatedIter->valid() && $literalIter->valid() ) {
 442                         $output .= $translatedIter->current();
 443                         $output .= $literalIter->current();
 444                         $translatedIter->next();
 445                         $literalIter->next();
 446                 }
 447
 448                 return $output;
 449         }
 450
 451         /**
 452          * Translate a string to a variant.
 453          * Doesn't parse rules or do any of that other stuff, for that use
 454          * convert() or convertTo().
 455          *
 456          * @param string $text Text to convert
 457          * @param string $variant Variant language code
 458          * @return string Translated text
 459          */
 460         public function translate( $text, $variant ) {
 461                 // If $text is empty or only includes spaces, do nothing
 462                 // Otherwise translate it
 463                 if ( trim( $text ) ) {
 464                         $this->loadTables();
 465                         $text = $this->mTables[$variant]->replace( $text );
 466                 }
 467                 return $text;
 468         }
 469
 470         /**
 471          * Call translate() to convert text to all valid variants.
 472          *
 473          * @param string $text The text to be converted
 474          * @return array Variant => converted text
 475          */
 476         public function autoConvertToAllVariants( $text ) {
 477                 $this->loadTables();
 478
 479                 $ret = [];
 480                 foreach ( $this->mVariants as $variant ) {
 481                         $ret[$variant] = $this->translate( $text, $variant );
 482                 }
 483
 484                 return $ret;
 485         }
 486
 487         /**
 488          * Apply manual conversion rules.
 489          *
 490          * @param ConverterRule $convRule
 491          */
 492         protected function applyManualConv( $convRule ) {
 493                 // Use syntax -{T|zh-cn:TitleCN; zh-tw:TitleTw}- to custom
 494                 // title conversion.
 495                 // T26072: $mConvRuleTitle was overwritten by other manual
 496                 // rule(s) not for title, this breaks the title conversion.
 497                 $newConvRuleTitle = $convRule->getTitle();
 498                 if ( $newConvRuleTitle ) {
 499                         // So I add an empty check for getTitle()
 500                         $this->mConvRuleTitle = $newConvRuleTitle;
 501                 }
 502
 503                 // merge/remove manual conversion rules to/from global table
 504                 $convTable = $convRule->getConvTable();
 505                 $action = $convRule->getRulesAction();
 506                 foreach ( $convTable as $variant => $pair ) {
 507                         if ( !$this->validateVariant( $variant ) ) {
 508                                 continue;
 509                         }
 510
 511                         if ( $action == 'add' ) {
 512                                 // More efficient than array_merge(), about 2.5 times.
 513                                 foreach ( $pair as $from => $to ) {
 514                                         $this->mTables[$variant]->setPair( $from, $to );
 515                                 }
 516                         } elseif ( $action == 'remove' ) {
 517                                 $this->mTables[$variant]->removeArray( $pair );
 518                         }
 519                 }
 520         }
 521
 522         /**
 523          * Auto convert a Title object to a readable string in the
 524          * preferred variant.
 525          *
 526          * @param Title $title A object of Title
 527          * @return string Converted title text
 528          */
 529         public function convertTitle( $title ) {
 530                 $variant = $this->getPreferredVariant();
 531                 $index = $title->getNamespace();
 532                 if ( $index !== NS_MAIN ) {
 533                         $text = $this->convertNamespace( $index, $variant ) . ':';
 534                 } else {
 535                         $text = '';
 536                 }
 537                 $text .= $this->translate( $title->getText(), $variant );
 538                 return $text;
 539         }
 540
 541         /**
 542          * Get the namespace display name in the preferred variant.
 543          *
 544          * @param int $index Namespace id
 545          * @param string|null $variant Variant code or null for preferred variant
 546          * @return string Namespace name for display
 547          */
 548         public function convertNamespace( $index, $variant = null ) {
 549                 if ( $index === NS_MAIN ) {
 550                         return '';
 551                 }
 552
 553                 if ( $variant === null ) {
 554                         $variant = $this->getPreferredVariant();
 555                 }
 556
 557                 $cache = MediaWikiServices::getInstance()->getLocalServerObjectCache();
 558                 $key = $cache->makeKey( 'languageconverter', 'namespace-text', $index, $variant );
 559                 $nsVariantText = $cache->get( $key );
 560                 if ( $nsVariantText !== false ) {
 561                         return $nsVariantText;
 562                 }
 563
 564                 // First check if a message gives a converted name in the target variant.
 565                 $nsConvMsg = wfMessage( 'conversion-ns' . $index )->inLanguage( $variant );
 566                 if ( $nsConvMsg->exists() ) {
 567                         $nsVariantText = $nsConvMsg->plain();
 568                 }
 569
 570                 // Then check if a message gives a converted name in content language
 571                 // which needs extra translation to the target variant.
 572                 if ( $nsVariantText === false ) {
 573                         $nsConvMsg = wfMessage( 'conversion-ns' . $index )->inContentLanguage();
 574                         if ( $nsConvMsg->exists() ) {
 575                                 $nsVariantText = $this->translate( $nsConvMsg->plain(), $variant );
 576                         }
 577                 }
 578
 579                 if ( $nsVariantText === false ) {
 580                         // No message exists, retrieve it from the target variant's namespace names.
 581                         $langObj = $this->mLangObj->factory( $variant );
 582                         $nsVariantText = $langObj->getFormattedNsText( $index );
 583                 }
 584
 585                 $cache->set( $key, $nsVariantText, 60 );
 586
 587                 return $nsVariantText;
 588         }
 589
 590         /**
 591          * Convert text to different variants of a language. The automatic
 592          * conversion is done in autoConvert(). Here we parse the text
 593          * marked with -{}-, which specifies special conversions of the
 594          * text that can not be accomplished in autoConvert().
 595          *
 596          * Syntax of the markup:
 597          * -{code1:text1;code2:text2;...}-  or
 598          * -{flags|code1:text1;code2:text2;...}-  or
 599          * -{text}- in which case no conversion should take place for text
 600          *
 601          * @param string $text Text to be converted
 602          * @return string Converted text
 603          */
 604         public function convert( $text ) {
 605                 $variant = $this->getPreferredVariant();
 606                 return $this->convertTo( $text, $variant );
 607         }
 608
 609         /**
 610          * Same as convert() except a extra parameter to custom variant.
 611          *
 612          * @param string $text Text to be converted
 613          * @param string $variant The target variant code
 614          * @return string Converted text
 615          */
 616         public function convertTo( $text, $variant ) {
 617                 global $wgDisableLangConversion;
 618                 if ( $wgDisableLangConversion ) {
 619                         return $text;
 620                 }
 621                 // Reset converter state for a new converter run.
 622                 $this->mConvRuleTitle = false;
 623                 return $this->recursiveConvertTopLevel( $text, $variant );
 624         }
 625
 626         /**
 627          * Recursively convert text on the outside. Allow to use nested
 628          * markups to custom rules.
 629          *
 630          * @param string $text Text to be converted
 631          * @param string $variant The target variant code
 632          * @param int $depth Depth of recursion
 633          * @return string Converted text
 634          */
 635         protected function recursiveConvertTopLevel( $text, $variant, $depth = 0 ) {
 636                 $startPos = 0;
 637                 $out = '';
 638                 $length = strlen( $text );
 639                 $shouldConvert = !$this->guessVariant( $text, $variant );
 640
 641                 while ( $startPos < $length ) {
 642                         $pos = strpos( $text, '-{', $startPos );
 643
 644                         if ( $pos === false ) {
 645                                 // No more markup, append final segment
 646                                 $fragment = substr( $text, $startPos );
 647                                 $out .= $shouldConvert ? $this->autoConvert( $fragment, $variant ) : $fragment;
 648                                 return $out;
 649                         }
 650
 651                         // Markup found
 652                         // Append initial segment
 653                         $fragment = substr( $text, $startPos, $pos - $startPos );
 654                         $out .= $shouldConvert ? $this->autoConvert( $fragment, $variant ) : $fragment;
 655
 656                         // Advance position
 657                         $startPos = $pos;
 658
 659                         // Do recursive conversion
 660                         $out .= $this->recursiveConvertRule( $text, $variant, $startPos, $depth + 1 );
 661                 }
 662
 663                 return $out;
 664         }
 665
 666         /**
 667          * Recursively convert text on the inside.
 668          *
 669          * @param string $text Text to be converted
 670          * @param string $variant The target variant code
 671          * @param int $startPos
 672          * @param int $depth Depth of recursion
 673          *
 674          * @throws MWException
 675          * @return string Converted text
 676          */
 677         protected function recursiveConvertRule( $text, $variant, &$startPos, $depth = 0 ) {
 678                 // Quick sanity check (no function calls)
 679                 if ( $text[$startPos] !== '-' || $text[$startPos + 1] !== '{' ) {
 680                         throw new MWException( __METHOD__ . ': invalid input string' );
 681                 }
 682
 683                 $startPos += 2;
 684                 $inner = '';
 685                 $warningDone = false;
 686                 $length = strlen( $text );
 687
 688                 while ( $startPos < $length ) {
 689                         $m = false;
 690                         preg_match( '/-\{|\}-/', $text, $m, PREG_OFFSET_CAPTURE, $startPos );
 691                         if ( !$m ) {
 692                                 // Unclosed rule
 693                                 break;
 694                         }
 695
 696                         $token = $m[0][0];
 697                         $pos = $m[0][1];
 698
 699                         // Markup found
 700                         // Append initial segment
 701                         $inner .= substr( $text, $startPos, $pos - $startPos );
 702
 703                         // Advance position
 704                         $startPos = $pos;
 705
 706                         switch ( $token ) {
 707                                 case '-{':
 708                                         // Check max depth
 709                                         if ( $depth >= $this->mMaxDepth ) {
 710                                                 $inner .= '-{';
 711                                                 if ( !$warningDone ) {
 712                                                         $inner .= '<span class="error">' .
 713                                                                 wfMessage( 'language-converter-depth-warning' )
 714                                                                         ->numParams( $this->mMaxDepth )->inContentLanguage()->text() .
 715                                                                 '</span>';
 716                                                         $warningDone = true;
 717                                                 }
 718                                                 $startPos += 2;
 719                                                 continue;
 720                                         }
 721                                         // Recursively parse another rule
 722                                         $inner .= $this->recursiveConvertRule( $text, $variant, $startPos, $depth + 1 );
 723                                         break;
 724                                 case '}-':
 725                                         // Apply the rule
 726                                         $startPos += 2;
 727                                         $rule = new ConverterRule( $inner, $this );
 728                                         $rule->parse( $variant );
 729                                         $this->applyManualConv( $rule );
 730                                         return $rule->getDisplay();
 731                                 default:
 732                                         throw new MWException( __METHOD__ . ': invalid regex match' );
 733                         }
 734                 }
 735
 736                 // Unclosed rule
 737                 if ( $startPos < $length ) {
 738                         $inner .= substr( $text, $startPos );
 739                 }
 740                 $startPos = $length;
 741                 return '-{' . $this->autoConvert( $inner, $variant );
 742         }
 743
 744         /**
 745          * If a language supports multiple variants, it is possible that
 746          * non-existing link in one variant actually exists in another variant.
 747          * This function tries to find it. See e.g. LanguageZh.php
 748          * The input parameters may be modified upon return
 749          *
 750          * @param string &$link The name of the link
 751          * @param Title &$nt The title object of the link
 752          * @param bool $ignoreOtherCond To disable other conditions when
 753          *   we need to transclude a template or update a category's link
 754          */
 755         public function findVariantLink( &$link, &$nt, $ignoreOtherCond = false ) {
 756                 # If the article has already existed, there is no need to
 757                 # check it again, otherwise it may cause a fault.
 758                 if ( is_object( $nt ) && $nt->exists() ) {
 759                         return;
 760                 }
 761
 762                 global $wgDisableLangConversion, $wgDisableTitleConversion, $wgRequest;
 763                 $isredir = $wgRequest->getText( 'redirect', 'yes' );
 764                 $action = $wgRequest->getText( 'action' );
 765                 if ( $action == 'edit' && $wgRequest->getBool( 'redlink' ) ) {
 766                         $action = 'view';
 767                 }
 768                 $linkconvert = $wgRequest->getText( 'linkconvert', 'yes' );
 769                 $disableLinkConversion = $wgDisableLangConversion
 770                         || $wgDisableTitleConversion;
 771                 $linkBatch = new LinkBatch();
 772
 773                 $ns = NS_MAIN;
 774
 775                 if ( $disableLinkConversion ||
 776                         ( !$ignoreOtherCond &&
 777                                 ( $isredir == 'no'
 778                                         || $action == 'edit'
 779                                         || $action == 'submit'
 780                                         || $linkconvert == 'no' ) ) ) {
 781                         return;
 782                 }
 783
 784                 if ( is_object( $nt ) ) {
 785                         $ns = $nt->getNamespace();
 786                 }
 787
 788                 $variants = $this->autoConvertToAllVariants( $link );
 789                 if ( !$variants ) { // give up
 790                         return;
 791                 }
 792
 793                 $titles = [];
 794
 795                 foreach ( $variants as $v ) {
 796                         if ( $v != $link ) {
 797                                 $varnt = Title::newFromText( $v, $ns );
 798                                 if ( !is_null( $varnt ) ) {
 799                                         $linkBatch->addObj( $varnt );
 800                                         $titles[] = $varnt;
 801                                 }
 802                         }
 803                 }
 804
 805                 // fetch all variants in single query
 806                 $linkBatch->execute();
 807
 808                 foreach ( $titles as $varnt ) {
 809                         if ( $varnt->getArticleID() > 0 ) {
 810                                 $nt = $varnt;
 811                                 $link = $varnt->getText();
 812                                 break;
 813                         }
 814                 }
 815         }
 816
 817         /**
 818          * Returns language specific hash options.
 819          *
 820          * @return string
 821          */
 822         public function getExtraHashOptions() {
 823                 $variant = $this->getPreferredVariant();
 824
 825                 return '!' . $variant;
 826         }
 827
 828         /**
 829          * Guess if a text is written in a variant. This should be implemented in subclasses.
 830          *
 831          * @param string $text The text to be checked
 832          * @param string $variant Language code of the variant to be checked for
 833          * @return bool True if $text appears to be written in $variant, false if not
 834          *
 835          * @author Nikola Smolenski <smolensk@eunet.rs>
 836          * @since 1.19
 837          */
 838         public function guessVariant( $text, $variant ) {
 839                 return false;
 840         }
 841
 842         /**
 843          * Load default conversion tables.
 844          * This method must be implemented in derived class.
 845          *
 846          * @private
 847          * @throws MWException
 848          */
 849         function loadDefaultTables() {
 850                 $class = static::class;
 851                 throw new MWException( "Must implement loadDefaultTables() method in class $class" );
 852         }
 853
 854         /**
 855          * Load conversion tables either from the cache or the disk.
 856          * @private
 857          * @param bool $fromCache Load from memcached? Defaults to true.
 858          */
 859         function loadTables( $fromCache = true ) {
 860                 global $wgLanguageConverterCacheType;
 861
 862                 if ( $this->mTablesLoaded ) {
 863                         return;
 864                 }
 865
 866                 $this->mTablesLoaded = true;
 867                 $this->mTables = false;
 868                 $cache = ObjectCache::getInstance( $wgLanguageConverterCacheType );
 869                 if ( $fromCache ) {
 870                         $this->mTables = $cache->get( $this->mCacheKey );
 871                 }
 872                 if ( !$this->mTables || !array_key_exists( self::CACHE_VERSION_KEY, $this->mTables ) ) {
 873                         // not in cache, or we need a fresh reload.
 874                         // We will first load the default tables
 875                         // then update them using things in MediaWiki:Conversiontable/*
 876                         $this->loadDefaultTables();
 877                         foreach ( $this->mVariants as $var ) {
 878                                 $cached = $this->parseCachedTable( $var );
 879                                 $this->mTables[$var]->mergeArray( $cached );
 880                         }
 881
 882                         $this->postLoadTables();
 883                         $this->mTables[self::CACHE_VERSION_KEY] = true;
 884
 885                         $cache->set( $this->mCacheKey, $this->mTables, 43200 );
 886                 }
 887         }
 888
 889         /**
 890          * Hook for post processing after conversion tables are loaded.
 891          */
 892         function postLoadTables() {
 893         }
 894
 895         /**
 896          * Reload the conversion tables.
 897          *
 898          * @private
 899          */
 900         function reloadTables() {
 901                 if ( $this->mTables ) {
 902                         unset( $this->mTables );
 903                 }
 904
 905                 $this->mTablesLoaded = false;
 906                 $this->loadTables( false );
 907         }
 908
 909         /**
 910          * Parse the conversion table stored in the cache.
 911          *
 912          * The tables should be in blocks of the following form:
 913          *              -{
 914          *                      word => word ;
 915          *                      word => word ;
 916          *                      ...
 917          *              }-
 918          *
 919          * To make the tables more manageable, subpages are allowed
 920          * and will be parsed recursively if $recursive == true.
 921          *
 922          * @param string $code Language code
 923          * @param string $subpage Subpage name
 924          * @param bool $recursive Parse subpages recursively? Defaults to true.
 925          *
 926          * @return array
 927          */
 928         function parseCachedTable( $code, $subpage = '', $recursive = true ) {
 929                 static $parsed = [];
 930
 931                 $key = 'Conversiontable/' . $code;
 932                 if ( $subpage ) {
 933                         $key .= '/' . $subpage;
 934                 }
 935                 if ( array_key_exists( $key, $parsed ) ) {
 936                         return [];
 937                 }
 938
 939                 $parsed[$key] = true;
 940
 941                 if ( $subpage === '' ) {
 942                         $txt = MessageCache::singleton()->getMsgFromNamespace( $key, $code );
 943                 } else {
 944                         $txt = false;
 945                         $title = Title::makeTitleSafe( NS_MEDIAWIKI, $key );
 946                         if ( $title && $title->exists() ) {
 947                                 $revision = Revision::newFromTitle( $title );
 948                                 if ( $revision ) {
 949                                         if ( $revision->getContentModel() == CONTENT_MODEL_WIKITEXT ) {
 950                                                 $txt = $revision->getContent( Revision::RAW )->getNativeData();
 951                                         }
 952
 953                                         // @todo in the future, use a specialized content model, perhaps based on json!
 954                                 }
 955                         }
 956                 }
 957
 958                 # Nothing to parse if there's no text
 959                 if ( $txt === false || $txt === null || $txt === '' ) {
 960                         return [];
 961                 }
 962
 963                 // get all subpage links of the form
 964                 // [[MediaWiki:Conversiontable/zh-xx/...|...]]
 965                 $linkhead = $this->mLangObj->getNsText( NS_MEDIAWIKI ) .
 966                         ':Conversiontable';
 967                 $subs = StringUtils::explode( '[[', $txt );
 968                 $sublinks = [];
 969                 foreach ( $subs as $sub ) {
 970                         $link = explode( ']]', $sub, 2 );
 971                         if ( count( $link ) != 2 ) {
 972                                 continue;
 973                         }
 974                         $b = explode( '|', $link[0], 2 );
 975                         $b = explode( '/', trim( $b[0] ), 3 );
 976                         if ( count( $b ) == 3 ) {
 977                                 $sublink = $b[2];
 978                         } else {
 979                                 $sublink = '';
 980                         }
 981
 982                         if ( $b[0] == $linkhead && $b[1] == $code ) {
 983                                 $sublinks[] = $sublink;
 984                         }
 985                 }
 986
 987                 // parse the mappings in this page
 988                 $blocks = StringUtils::explode( '-{', $txt );
 989                 $ret = [];
 990                 $first = true;
 991                 foreach ( $blocks as $block ) {
 992                         if ( $first ) {
 993                                 // Skip the part before the first -{
 994                                 $first = false;
 995                                 continue;
 996                         }
 997                         $mappings = explode( '}-', $block, 2 )[0];
 998                         $stripped = str_replace( [ "'", '"', '*', '#' ], '', $mappings );
 999                         $table = StringUtils::explode( ';', $stripped );
1000                         foreach ( $table as $t ) {
1001                                 $m = explode( '=>', $t, 3 );
1002                                 if ( count( $m ) != 2 ) {
1003                                         continue;
1004                                 }
1005                                 // trim any trailling comments starting with '//'
1006                                 $tt = explode( '//', $m[1], 2 );
1007                                 $ret[trim( $m[0] )] = trim( $tt[0] );
1008                         }
1009                 }
1010
1011                 // recursively parse the subpages
1012                 if ( $recursive ) {
1013                         foreach ( $sublinks as $link ) {
1014                                 $s = $this->parseCachedTable( $code, $link, $recursive );
1015                                 $ret = $s + $ret;
1016                         }
1017                 }
1018
1019                 if ( $this->mUcfirst ) {
1020                         foreach ( $ret as $k => $v ) {
1021                                 $ret[$this->mLangObj->ucfirst( $k )] = $this->mLangObj->ucfirst( $v );
1022                         }
1023                 }
1024                 return $ret;
1025         }
1026
1027         /**
1028          * Enclose a string with the "no conversion" tag. This is used by
1029          * various functions in the Parser.
1030          *
1031          * @param string $text Text to be tagged for no conversion
1032          * @param bool $noParse Unused
1033          * @return string The tagged text
1034          */
1035         public function markNoConversion( $text, $noParse = false ) {
1036                 # don't mark if already marked
1037                 if ( strpos( $text, '-{' ) || strpos( $text, '}-' ) ) {
1038                         return $text;
1039                 }
1040
1041                 $ret = "-{R|$text}-";
1042                 return $ret;
1043         }
1044
1045         /**
1046          * Convert the sorting key for category links. This should make different
1047          * keys that are variants of each other map to the same key.
1048          *
1049          * @param string $key
1050          *
1051          * @return string
1052          */
1053         function convertCategoryKey( $key ) {
1054                 return $key;
1055         }
1056
1057         /**
1058          * Refresh the cache of conversion tables when
1059          * MediaWiki:Conversiontable* is updated.
1060          *
1061          * @param Title $titleobj The Title of the page being updated
1062          */
1063         public function updateConversionTable( Title $titleobj ) {
1064                 if ( $titleobj->getNamespace() == NS_MEDIAWIKI ) {
1065                         $title = $titleobj->getDBkey();
1066                         $t = explode( '/', $title, 3 );
1067                         $c = count( $t );
1068                         if ( $c > 1 && $t[0] == 'Conversiontable' ) {
1069                                 if ( $this->validateVariant( $t[1] ) ) {
1070                                         $this->reloadTables();
1071                                 }
1072                         }
1073                 }
1074         }
1075
1076         /**
1077          * Get the cached separator pattern for ConverterRule::parseRules()
1078          * @return string
1079          */
1080         function getVarSeparatorPattern() {
1081                 if ( is_null( $this->mVarSeparatorPattern ) ) {
1082                         // varsep_pattern for preg_split:
1083                         // text should be splited by ";" only if a valid variant
1084                         // name exist after the markup, for example:
1085                         //  -{zh-hans:<span style="font-size:120%;">xxx</span>;zh-hant:\
1086                         //  <span style="font-size:120%;">yyy</span>;}-
1087                         // we should split it as:
1088                         //  [
1089                         //    [0] => 'zh-hans:<span style="font-size:120%;">xxx</span>'
1090                         //    [1] => 'zh-hant:<span style="font-size:120%;">yyy</span>'
1091                         //    [2] => ''
1092                         //  ]
1093                         $pat = '/;\s*(?=';
1094                         foreach ( $this->mVariants as $variant ) {
1095                                 // zh-hans:xxx;zh-hant:yyy
1096                                 $pat .= $variant . '\s*:|';
1097                                 // xxx=>zh-hans:yyy; xxx=>zh-hant:zzz
1098                                 $pat .= '[^;]*?=>\s*' . $variant . '\s*:|';
1099                         }
1100                         $pat .= '\s*$)/';
1101                         $this->mVarSeparatorPattern = $pat;
1102                 }
1103                 return $this->mVarSeparatorPattern;
1104         }
1105 }