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