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 use MediaWiki\Logger\LoggerFactory;
  24
  25 /**
  26  * Base class for language conversion.
  27  * @ingroup Language
  28  *
  29  * @author Zhengzhu Feng <zhengzhu@gmail.com>
  30  * @author fdcn <fdcn64@gmail.com>
  31  * @author shinjiman <shinjiman@gmail.com>
  32  * @author PhiLiP <philip.npc@gmail.com>
  33  */
  34 class LanguageConverter {
  35         /**
  36          * languages supporting variants
  37          * @since 1.20
  38          * @var array
  39          */
  40         static public $languagesWithVariants = [
  41                 'en',
  42                 'gan',
  43                 'iu',
  44                 'kk',
  45                 'ku',
  46                 'shi',
  47                 'sr',
  48                 'tg',
  49                 'uz',
  50                 'zh',
  51         ];
  52
  53         public $mMainLanguageCode;
  54
  55         /**
  56          * @var string[]
  57          */
  58         public $mVariants;
  59         public $mVariantFallbacks;
  60         public $mVariantNames;
  61         public $mTablesLoaded = false;
  62         public $mTables;
  63         // 'bidirectional' 'unidirectional' 'disable' for each variant
  64         public $mManualLevel;
  65
  66         public $mLangObj;
  67         public $mFlags;
  68         public $mDescCodeSep = ':', $mDescVarSep = ';';
  69         public $mUcfirst = false;
  70         public $mConvRuleTitle = false;
  71         public $mURLVariant;
  72         public $mUserVariant;
  73         public $mHeaderVariant;
  74         public $mMaxDepth = 10;
  75         public $mVarSeparatorPattern;
  76
  77         const CACHE_VERSION_KEY = 'VERSION 7';
  78
  79         /**
  80          * @param Language $langobj
  81          * @param string $maincode The main language code of this language
  82          * @param string[] $variants The supported variants of this language
  83          * @param array $variantfallbacks The fallback language of each variant
  84          * @param array $flags Defining the custom strings that maps to the flags
  85          * @param array $manualLevel Limit for supported variants
  86          */
  87         public function __construct( $langobj, $maincode, $variants = [],
  88                                                                 $variantfallbacks = [], $flags = [],
  89                                                                 $manualLevel = [] ) {
  90                 global $wgDisabledVariants;
  91                 $this->mLangObj = $langobj;
  92                 $this->mMainLanguageCode = $maincode;
  93                 $this->mVariants = array_diff( $variants, $wgDisabledVariants );
  94                 $this->mVariantFallbacks = $variantfallbacks;
  95                 $this->mVariantNames = Language::fetchLanguageNames();
  96                 $defaultflags = [
  97                         // 'S' show converted text
  98                         // '+' add rules for alltext
  99                         // 'E' the gave flags is error
 100                         // these flags above are reserved for program
 101                         'A' => 'A',   // add rule for convert code (all text convert)
 102                         'T' => 'T',   // title convert
 103                         'R' => 'R',   // raw content
 104                         'D' => 'D',   // convert description (subclass implement)
 105                         '-' => '-',   // remove convert (not implement)
 106                         'H' => 'H',   // add rule for convert code (but no display in placed code)
 107                         'N' => 'N',   // current variant name
 108                 ];
 109                 $this->mFlags = array_merge( $defaultflags, $flags );
 110                 foreach ( $this->mVariants as $v ) {
 111                         if ( array_key_exists( $v, $manualLevel ) ) {
 112                                 $this->mManualLevel[$v] = $manualLevel[$v];
 113                         } else {
 114                                 $this->mManualLevel[$v] = 'bidirectional';
 115                         }
 116                         $this->mFlags[$v] = $v;
 117                 }
 118         }
 119
 120         /**
 121          * Get all valid variants.
 122          * Call this instead of using $this->mVariants directly.
 123          *
 124          * @return string[] Contains all valid variants
 125          */
 126         public function getVariants() {
 127                 return $this->mVariants;
 128         }
 129
 130         /**
 131          * In case some variant is not defined in the markup, we need
 132          * to have some fallback. For example, in zh, normally people
 133          * will define zh-hans and zh-hant, but less so for zh-sg or zh-hk.
 134          * when zh-sg is preferred but not defined, we will pick zh-hans
 135          * in this case. Right now this is only used by zh.
 136          *
 137          * @param string $variant The language code of the variant
 138          * @return string|array The code of the fallback language or the
 139          *   main code if there is no fallback
 140          */
 141         public function getVariantFallbacks( $variant ) {
 142                 if ( isset( $this->mVariantFallbacks[$variant] ) ) {
 143                         return $this->mVariantFallbacks[$variant];
 144                 }
 145                 return $this->mMainLanguageCode;
 146         }
 147
 148         /**
 149          * Get the title produced by the conversion rule.
 150          * @return string The converted title text
 151          */
 152         public function getConvRuleTitle() {
 153                 return $this->mConvRuleTitle;
 154         }
 155
 156         /**
 157          * Get preferred language variant.
 158          * @return string The preferred language code
 159          */
 160         public function getPreferredVariant() {
 161                 global $wgDefaultLanguageVariant, $wgUser;
 162
 163                 $req = $this->getURLVariant();
 164
 165                 if ( $wgUser->isSafeToLoad() && $wgUser->isLoggedIn() && !$req ) {
 166                         $req = $this->getUserVariant();
 167                 } elseif ( !$req ) {
 168                         $req = $this->getHeaderVariant();
 169                 }
 170
 171                 if ( $wgDefaultLanguageVariant && !$req ) {
 172                         $req = $this->validateVariant( $wgDefaultLanguageVariant );
 173                 }
 174
 175                 // This function, unlike the other get*Variant functions, is
 176                 // not memoized (i.e. there return value is not cached) since
 177                 // new information might appear during processing after this
 178                 // is first called.
 179                 if ( $this->validateVariant( $req ) ) {
 180                         return $req;
 181                 }
 182                 return $this->mMainLanguageCode;
 183         }
 184
 185         /**
 186          * Get default variant.
 187          * This function would not be affected by user's settings
 188          * @return string The default variant code
 189          */
 190         public function getDefaultVariant() {
 191                 global $wgDefaultLanguageVariant;
 192
 193                 $req = $this->getURLVariant();
 194
 195                 if ( !$req ) {
 196                         $req = $this->getHeaderVariant();
 197                 }
 198
 199                 if ( $wgDefaultLanguageVariant && !$req ) {
 200                         $req = $this->validateVariant( $wgDefaultLanguageVariant );
 201                 }
 202
 203                 if ( $req ) {
 204                         return $req;
 205                 }
 206                 return $this->mMainLanguageCode;
 207         }
 208
 209         /**
 210          * Validate the variant
 211          * @param string $variant The variant to validate
 212          * @return mixed Returns the variant if it is valid, null otherwise
 213          */
 214         public function validateVariant( $variant = null ) {
 215                 if ( $variant !== null && in_array( $variant, $this->mVariants ) ) {
 216                         return $variant;
 217                 }
 218                 return null;
 219         }
 220
 221         /**
 222          * Get the variant specified in the URL
 223          *
 224          * @return mixed Variant if one found, false otherwise.
 225          */
 226         public function getURLVariant() {
 227                 global $wgRequest;
 228
 229                 if ( $this->mURLVariant ) {
 230                         return $this->mURLVariant;
 231                 }
 232
 233                 // see if the preference is set in the request
 234                 $ret = $wgRequest->getText( 'variant' );
 235
 236                 if ( !$ret ) {
 237                         $ret = $wgRequest->getVal( 'uselang' );
 238                 }
 239
 240                 $this->mURLVariant = $this->validateVariant( $ret );
 241                 return $this->mURLVariant;
 242         }
 243
 244         /**
 245          * Determine if the user has a variant set.
 246          *
 247          * @return mixed Variant if one found, false otherwise.
 248          */
 249         protected function getUserVariant() {
 250                 global $wgUser, $wgContLang;
 251
 252                 // memoizing this function wreaks havoc on parserTest.php
 253                 /*
 254                 if ( $this->mUserVariant ) {
 255                         return $this->mUserVariant;
 256                 }
 257                 */
 258
 259                 // Get language variant preference from logged in users
 260                 // Don't call this on stub objects because that causes infinite
 261                 // recursion during initialisation
 262                 if ( !$wgUser->isSafeToLoad() ) {
 263                         return false;
 264                 }
 265                 if ( $wgUser->isLoggedIn() ) {
 266                         if ( $this->mMainLanguageCode == $wgContLang->getCode() ) {
 267                                 $ret = $wgUser->getOption( 'variant' );
 268                         } else {
 269                                 $ret = $wgUser->getOption( 'variant-' . $this->mMainLanguageCode );
 270                         }
 271                 } else {
 272                         // figure out user lang without constructing wgLang to avoid
 273                         // infinite recursion
 274                         $ret = $wgUser->getOption( 'language' );
 275                 }
 276
 277                 $this->mUserVariant = $this->validateVariant( $ret );
 278                 return $this->mUserVariant;
 279         }
 280
 281         /**
 282          * Determine the language variant from the Accept-Language header.
 283          *
 284          * @return mixed Variant if one found, false otherwise.
 285          */
 286         protected function getHeaderVariant() {
 287                 global $wgRequest;
 288
 289                 if ( $this->mHeaderVariant ) {
 290                         return $this->mHeaderVariant;
 291                 }
 292
 293                 // see if some supported language variant is set in the
 294                 // HTTP header.
 295                 $languages = array_keys( $wgRequest->getAcceptLang() );
 296                 if ( empty( $languages ) ) {
 297                         return null;
 298                 }
 299
 300                 $fallbackLanguages = [];
 301                 foreach ( $languages as $language ) {
 302                         $this->mHeaderVariant = $this->validateVariant( $language );
 303                         if ( $this->mHeaderVariant ) {
 304                                 break;
 305                         }
 306
 307                         // To see if there are fallbacks of current language.
 308                         // We record these fallback variants, and process
 309                         // them later.
 310                         $fallbacks = $this->getVariantFallbacks( $language );
 311                         if ( is_string( $fallbacks ) && $fallbacks !== $this->mMainLanguageCode ) {
 312                                 $fallbackLanguages[] = $fallbacks;
 313                         } elseif ( is_array( $fallbacks ) ) {
 314                                 $fallbackLanguages =
 315                                         array_merge( $fallbackLanguages, $fallbacks );
 316                         }
 317                 }
 318
 319                 if ( !$this->mHeaderVariant ) {
 320                         // process fallback languages now
 321                         $fallback_languages = array_unique( $fallbackLanguages );
 322                         foreach ( $fallback_languages as $language ) {
 323                                 $this->mHeaderVariant = $this->validateVariant( $language );
 324                                 if ( $this->mHeaderVariant ) {
 325                                         break;
 326                                 }
 327                         }
 328                 }
 329
 330                 return $this->mHeaderVariant;
 331         }
 332
 333         /**
 334          * Dictionary-based conversion.
 335          * This function would not parse the conversion rules.
 336          * If you want to parse rules, try to use convert() or
 337          * convertTo().
 338          *
 339          * @param string $text The text to be converted
 340          * @param bool|string $toVariant The target language code
 341          * @return string The converted text
 342          */
 343         public function autoConvert( $text, $toVariant = false ) {
 344                 $this->loadTables();
 345
 346                 if ( !$toVariant ) {
 347                         $toVariant = $this->getPreferredVariant();
 348                         if ( !$toVariant ) {
 349                                 return $text;
 350                         }
 351                 }
 352
 353                 if ( $this->guessVariant( $text, $toVariant ) ) {
 354                         return $text;
 355                 }
 356
 357                 /* we convert everything except:
 358                    1. HTML markups (anything between < and >)
 359                    2. HTML entities
 360                    3. placeholders created by the parser
 361                    IMPORTANT: Beware of failure from pcre.backtrack_limit (T124404).
 362                    Minimize use of backtracking where possible.
 363                 */
 364                 $marker = '|' . Parser::MARKER_PREFIX . '[^\x7f]++\x7f';
 365
 366                 // this one is needed when the text is inside an HTML markup
 367                 $htmlfix = '|<[^>\004]++(?=\004$)|^[^<>]*+>';
 368
 369                 // Optimize for the common case where these tags have
 370                 // few or no children. Thus try and possesively get as much as
 371                 // possible, and only engage in backtracking when we hit a '<'.
 372
 373                 // disable convert to variants between <code> tags
 374                 $codefix = '<code>[^<]*+(?:(?:(?!<\/code>).)[^<]*+)*+<\/code>|';
 375                 // disable conversion of <script> tags
 376                 $scriptfix = '<script[^>]*+>[^<]*+(?:(?:(?!<\/script>).)[^<]*+)*+<\/script>|';
 377                 // disable conversion of <pre> tags
 378                 $prefix = '<pre[^>]*+>[^<]*+(?:(?:(?!<\/pre>).)[^<]*+)*+<\/pre>|';
 379
 380                 $reg = '/' . $codefix . $scriptfix . $prefix .
 381                         '<[^>]++>|&[a-zA-Z#][a-z0-9]++;' . $marker . $htmlfix . '|\004$/s';
 382                 $startPos = 0;
 383                 $sourceBlob = '';
 384                 $literalBlob = '';
 385
 386                 // Guard against delimiter nulls in the input
 387                 // (should never happen: see T159174)
 388                 $text = str_replace( "\000", '', $text );
 389
 390                 $markupMatches = null;
 391                 $elementMatches = null;
 392
 393                 // We add a marker (\004) at the end of text, to ensure we always match the
 394                 // entire text (Otherwise, pcre.backtrack_limit might cause silent failure)
 395                 while ( $startPos < strlen( $text ) ) {
 396                         if ( preg_match( $reg, $text . "\004", $markupMatches, PREG_OFFSET_CAPTURE, $startPos ) ) {
 397                                 $elementPos = $markupMatches[0][1];
 398                                 $element = $markupMatches[0][0];
 399                                 if ( $element === "\004" ) {
 400                                         // We hit the end.
 401                                         $elementPos = strlen( $text );
 402                                         $element = '';
 403                                 }
 404                         } else {
 405                                 // If we hit here, then Language Converter could be tricked
 406                                 // into doing an XSS, so we refuse to translate.
 407                                 // If non-crazy input manages to reach this code path,
 408                                 // we should consider it a bug.
 409                                 $log = LoggerFactory::getInstance( 'languageconverter' );
 410                                 $log->error( "Hit pcre.backtrack_limit in " . __METHOD__
 411                                         . ". Disabling language conversion for this page.",
 412                                         array(
 413                                                 "method" => __METHOD__,
 414                                                 "variant" => $toVariant,
 415                                                 "startOfText" => substr( $text, 0, 500 )
 416                                         )
 417                                 );
 418                                 return $text;
 419                         }
 420                         // Queue the part before the markup for translation in a batch
 421                         $sourceBlob .= substr( $text, $startPos, $elementPos - $startPos ) . "\000";
 422
 423                         // Advance to the next position
 424                         $startPos = $elementPos + strlen( $element );
 425
 426                         // Translate any alt or title attributes inside the matched element
 427                         if ( $element !== ''
 428                                 && preg_match( '/^(<[^>\s]*+)\s([^>]*+)(.*+)$/', $element, $elementMatches )
 429                         ) {
 430                                 $attrs = Sanitizer::decodeTagAttributes( $elementMatches[2] );
 431                                 $changed = false;
 432                                 foreach ( [ 'title', 'alt' ] as $attrName ) {
 433                                         if ( !isset( $attrs[$attrName] ) ) {
 434                                                 continue;
 435                                         }
 436                                         $attr = $attrs[$attrName];
 437                                         // Don't convert URLs
 438                                         if ( !strpos( $attr, '://' ) ) {
 439                                                 $attr = $this->recursiveConvertTopLevel( $attr, $toVariant );
 440                                         }
 441
 442                                         if ( $attr !== $attrs[$attrName] ) {
 443                                                 $attrs[$attrName] = $attr;
 444                                                 $changed = true;
 445                                         }
 446                                 }
 447                                 if ( $changed ) {
 448                                         $element = $elementMatches[1] . Html::expandAttributes( $attrs ) .
 449                                                 $elementMatches[3];
 450                                 }
 451                         }
 452                         $literalBlob .= $element . "\000";
 453                 }
 454
 455                 // Do the main translation batch
 456                 $translatedBlob = $this->translate( $sourceBlob, $toVariant );
 457
 458                 // Put the output back together
 459                 $translatedIter = StringUtils::explode( "\000", $translatedBlob );
 460                 $literalIter = StringUtils::explode( "\000", $literalBlob );
 461                 $output = '';
 462                 while ( $translatedIter->valid() && $literalIter->valid() ) {
 463                         $output .= $translatedIter->current();
 464                         $output .= $literalIter->current();
 465                         $translatedIter->next();
 466                         $literalIter->next();
 467                 }
 468
 469                 return $output;
 470         }
 471
 472         /**
 473          * Translate a string to a variant.
 474          * Doesn't parse rules or do any of that other stuff, for that use
 475          * convert() or convertTo().
 476          *
 477          * @param string $text Text to convert
 478          * @param string $variant Variant language code
 479          * @return string Translated text
 480          */
 481         public function translate( $text, $variant ) {
 482                 // If $text is empty or only includes spaces, do nothing
 483                 // Otherwise translate it
 484                 if ( trim( $text ) ) {
 485                         $this->loadTables();
 486                         $text = $this->mTables[$variant]->replace( $text );
 487                 }
 488                 return $text;
 489         }
 490
 491         /**
 492          * Call translate() to convert text to all valid variants.
 493          *
 494          * @param string $text The text to be converted
 495          * @return array Variant => converted text
 496          */
 497         public function autoConvertToAllVariants( $text ) {
 498                 $this->loadTables();
 499
 500                 $ret = [];
 501                 foreach ( $this->mVariants as $variant ) {
 502                         $ret[$variant] = $this->translate( $text, $variant );
 503                 }
 504
 505                 return $ret;
 506         }
 507
 508         /**
 509          * Apply manual conversion rules.
 510          *
 511          * @param ConverterRule $convRule
 512          */
 513         protected function applyManualConv( $convRule ) {
 514                 // Use syntax -{T|zh-cn:TitleCN; zh-tw:TitleTw}- to custom
 515                 // title conversion.
 516                 // T26072: $mConvRuleTitle was overwritten by other manual
 517                 // rule(s) not for title, this breaks the title conversion.
 518                 $newConvRuleTitle = $convRule->getTitle();
 519                 if ( $newConvRuleTitle ) {
 520                         // So I add an empty check for getTitle()
 521                         $this->mConvRuleTitle = $newConvRuleTitle;
 522                 }
 523
 524                 // merge/remove manual conversion rules to/from global table
 525                 $convTable = $convRule->getConvTable();
 526                 $action = $convRule->getRulesAction();
 527                 foreach ( $convTable as $variant => $pair ) {
 528                         if ( !$this->validateVariant( $variant ) ) {
 529                                 continue;
 530                         }
 531
 532                         if ( $action == 'add' ) {
 533                                 // More efficient than array_merge(), about 2.5 times.
 534                                 foreach ( $pair as $from => $to ) {
 535                                         $this->mTables[$variant]->setPair( $from, $to );
 536                                 }
 537                         } elseif ( $action == 'remove' ) {
 538                                 $this->mTables[$variant]->removeArray( $pair );
 539                         }
 540                 }
 541         }
 542
 543         /**
 544          * Auto convert a Title object to a readable string in the
 545          * preferred variant.
 546          *
 547          * @param Title $title A object of Title
 548          * @return string Converted title text
 549          */
 550         public function convertTitle( $title ) {
 551                 $variant = $this->getPreferredVariant();
 552                 $index = $title->getNamespace();
 553                 if ( $index !== NS_MAIN ) {
 554                         $text = $this->convertNamespace( $index, $variant ) . ':';
 555                 } else {
 556                         $text = '';
 557                 }
 558                 $text .= $this->translate( $title->getText(), $variant );
 559                 return $text;
 560         }
 561
 562         /**
 563          * Get the namespace display name in the preferred variant.
 564          *
 565          * @param int $index Namespace id
 566          * @param string|null $variant Variant code or null for preferred variant
 567          * @return string Namespace name for display
 568          */
 569         public function convertNamespace( $index, $variant = null ) {
 570                 if ( $index === NS_MAIN ) {
 571                         return '';
 572                 }
 573
 574                 if ( $variant === null ) {
 575                         $variant = $this->getPreferredVariant();
 576                 }
 577
 578                 $cache = MediaWikiServices::getInstance()->getLocalServerObjectCache();
 579                 $key = $cache->makeKey( 'languageconverter', 'namespace-text', $index, $variant );
 580                 $nsVariantText = $cache->get( $key );
 581                 if ( $nsVariantText !== false ) {
 582                         return $nsVariantText;
 583                 }
 584
 585                 // First check if a message gives a converted name in the target variant.
 586                 $nsConvMsg = wfMessage( 'conversion-ns' . $index )->inLanguage( $variant );
 587                 if ( $nsConvMsg->exists() ) {
 588                         $nsVariantText = $nsConvMsg->plain();
 589                 }
 590
 591                 // Then check if a message gives a converted name in content language
 592                 // which needs extra translation to the target variant.
 593                 if ( $nsVariantText === false ) {
 594                         $nsConvMsg = wfMessage( 'conversion-ns' . $index )->inContentLanguage();
 595                         if ( $nsConvMsg->exists() ) {
 596                                 $nsVariantText = $this->translate( $nsConvMsg->plain(), $variant );
 597                         }
 598                 }
 599
 600                 if ( $nsVariantText === false ) {
 601                         // No message exists, retrieve it from the target variant's namespace names.
 602                         $langObj = $this->mLangObj->factory( $variant );
 603                         $nsVariantText = $langObj->getFormattedNsText( $index );
 604                 }
 605
 606                 $cache->set( $key, $nsVariantText, 60 );
 607
 608                 return $nsVariantText;
 609         }
 610
 611         /**
 612          * Convert text to different variants of a language. The automatic
 613          * conversion is done in autoConvert(). Here we parse the text
 614          * marked with -{}-, which specifies special conversions of the
 615          * text that can not be accomplished in autoConvert().
 616          *
 617          * Syntax of the markup:
 618          * -{code1:text1;code2:text2;...}-  or
 619          * -{flags|code1:text1;code2:text2;...}-  or
 620          * -{text}- in which case no conversion should take place for text
 621          *
 622          * @param string $text Text to be converted
 623          * @return string Converted text
 624          */
 625         public function convert( $text ) {
 626                 $variant = $this->getPreferredVariant();
 627                 return $this->convertTo( $text, $variant );
 628         }
 629
 630         /**
 631          * Same as convert() except a extra parameter to custom variant.
 632          *
 633          * @param string $text Text to be converted
 634          * @param string $variant The target variant code
 635          * @return string Converted text
 636          */
 637         public function convertTo( $text, $variant ) {
 638                 global $wgDisableLangConversion;
 639                 if ( $wgDisableLangConversion ) {
 640                         return $text;
 641                 }
 642                 // Reset converter state for a new converter run.
 643                 $this->mConvRuleTitle = false;
 644                 return $this->recursiveConvertTopLevel( $text, $variant );
 645         }
 646
 647         /**
 648          * Recursively convert text on the outside. Allow to use nested
 649          * markups to custom rules.
 650          *
 651          * @param string $text Text to be converted
 652          * @param string $variant The target variant code
 653          * @param int $depth Depth of recursion
 654          * @return string Converted text
 655          */
 656         protected function recursiveConvertTopLevel( $text, $variant, $depth = 0 ) {
 657                 $startPos = 0;
 658                 $out = '';
 659                 $length = strlen( $text );
 660                 $shouldConvert = !$this->guessVariant( $text, $variant );
 661
 662                 while ( $startPos < $length ) {
 663                         $pos = strpos( $text, '-{', $startPos );
 664
 665                         if ( $pos === false ) {
 666                                 // No more markup, append final segment
 667                                 $fragment = substr( $text, $startPos );
 668                                 $out .= $shouldConvert ? $this->autoConvert( $fragment, $variant ) : $fragment;
 669                                 return $out;
 670                         }
 671
 672                         // Markup found
 673                         // Append initial segment
 674                         $fragment = substr( $text, $startPos, $pos - $startPos );
 675                         $out .= $shouldConvert ? $this->autoConvert( $fragment, $variant ) : $fragment;
 676
 677                         // Advance position
 678                         $startPos = $pos;
 679
 680                         // Do recursive conversion
 681                         $out .= $this->recursiveConvertRule( $text, $variant, $startPos, $depth + 1 );
 682                 }
 683
 684                 return $out;
 685         }
 686
 687         /**
 688          * Recursively convert text on the inside.
 689          *
 690          * @param string $text Text to be converted
 691          * @param string $variant The target variant code
 692          * @param int &$startPos
 693          * @param int $depth Depth of recursion
 694          *
 695          * @throws MWException
 696          * @return string Converted text
 697          */
 698         protected function recursiveConvertRule( $text, $variant, &$startPos, $depth = 0 ) {
 699                 // Quick sanity check (no function calls)
 700                 if ( $text[$startPos] !== '-' || $text[$startPos + 1] !== '{' ) {
 701                         throw new MWException( __METHOD__ . ': invalid input string' );
 702                 }
 703
 704                 $startPos += 2;
 705                 $inner = '';
 706                 $warningDone = false;
 707                 $length = strlen( $text );
 708
 709                 while ( $startPos < $length ) {
 710                         $m = false;
 711                         preg_match( '/-\{|\}-/', $text, $m, PREG_OFFSET_CAPTURE, $startPos );
 712                         if ( !$m ) {
 713                                 // Unclosed rule
 714                                 break;
 715                         }
 716
 717                         $token = $m[0][0];
 718                         $pos = $m[0][1];
 719
 720                         // Markup found
 721                         // Append initial segment
 722                         $inner .= substr( $text, $startPos, $pos - $startPos );
 723
 724                         // Advance position
 725                         $startPos = $pos;
 726
 727                         switch ( $token ) {
 728                                 case '-{':
 729                                         // Check max depth
 730                                         if ( $depth >= $this->mMaxDepth ) {
 731                                                 $inner .= '-{';
 732                                                 if ( !$warningDone ) {
 733                                                         $inner .= '<span class="error">' .
 734                                                                 wfMessage( 'language-converter-depth-warning' )
 735                                                                         ->numParams( $this->mMaxDepth )->inContentLanguage()->text() .
 736                                                                 '</span>';
 737                                                         $warningDone = true;
 738                                                 }
 739                                                 $startPos += 2;
 740                                                 continue;
 741                                         }
 742                                         // Recursively parse another rule
 743                                         $inner .= $this->recursiveConvertRule( $text, $variant, $startPos, $depth + 1 );
 744                                         break;
 745                                 case '}-':
 746                                         // Apply the rule
 747                                         $startPos += 2;
 748                                         $rule = new ConverterRule( $inner, $this );
 749                                         $rule->parse( $variant );
 750                                         $this->applyManualConv( $rule );
 751                                         return $rule->getDisplay();
 752                                 default:
 753                                         throw new MWException( __METHOD__ . ': invalid regex match' );
 754                         }
 755                 }
 756
 757                 // Unclosed rule
 758                 if ( $startPos < $length ) {
 759                         $inner .= substr( $text, $startPos );
 760                 }
 761                 $startPos = $length;
 762                 return '-{' . $this->autoConvert( $inner, $variant );
 763         }
 764
 765         /**
 766          * If a language supports multiple variants, it is possible that
 767          * non-existing link in one variant actually exists in another variant.
 768          * This function tries to find it. See e.g. LanguageZh.php
 769          * The input parameters may be modified upon return
 770          *
 771          * @param string &$link The name of the link
 772          * @param Title &$nt The title object of the link
 773          * @param bool $ignoreOtherCond To disable other conditions when
 774          *   we need to transclude a template or update a category's link
 775          */
 776         public function findVariantLink( &$link, &$nt, $ignoreOtherCond = false ) {
 777                 # If the article has already existed, there is no need to
 778                 # check it again, otherwise it may cause a fault.
 779                 if ( is_object( $nt ) && $nt->exists() ) {
 780                         return;
 781                 }
 782
 783                 global $wgDisableLangConversion, $wgDisableTitleConversion, $wgRequest;
 784                 $isredir = $wgRequest->getText( 'redirect', 'yes' );
 785                 $action = $wgRequest->getText( 'action' );
 786                 if ( $action == 'edit' && $wgRequest->getBool( 'redlink' ) ) {
 787                         $action = 'view';
 788                 }
 789                 $linkconvert = $wgRequest->getText( 'linkconvert', 'yes' );
 790                 $disableLinkConversion = $wgDisableLangConversion
 791                         || $wgDisableTitleConversion;
 792                 $linkBatch = new LinkBatch();
 793
 794                 $ns = NS_MAIN;
 795
 796                 if ( $disableLinkConversion ||
 797                         ( !$ignoreOtherCond &&
 798                                 ( $isredir == 'no'
 799                                         || $action == 'edit'
 800                                         || $action == 'submit'
 801                                         || $linkconvert == 'no' ) ) ) {
 802                         return;
 803                 }
 804
 805                 if ( is_object( $nt ) ) {
 806                         $ns = $nt->getNamespace();
 807                 }
 808
 809                 $variants = $this->autoConvertToAllVariants( $link );
 810                 if ( !$variants ) { // give up
 811                         return;
 812                 }
 813
 814                 $titles = [];
 815
 816                 foreach ( $variants as $v ) {
 817                         if ( $v != $link ) {
 818                                 $varnt = Title::newFromText( $v, $ns );
 819                                 if ( !is_null( $varnt ) ) {
 820                                         $linkBatch->addObj( $varnt );
 821                                         $titles[] = $varnt;
 822                                 }
 823                         }
 824                 }
 825
 826                 // fetch all variants in single query
 827                 $linkBatch->execute();
 828
 829                 foreach ( $titles as $varnt ) {
 830                         if ( $varnt->getArticleID() > 0 ) {
 831                                 $nt = $varnt;
 832                                 $link = $varnt->getText();
 833                                 break;
 834                         }
 835                 }
 836         }
 837
 838         /**
 839          * Returns language specific hash options.
 840          *
 841          * @return string
 842          */
 843         public function getExtraHashOptions() {
 844                 $variant = $this->getPreferredVariant();
 845
 846                 return '!' . $variant;
 847         }
 848
 849         /**
 850          * Guess if a text is written in a variant. This should be implemented in subclasses.
 851          *
 852          * @param string $text The text to be checked
 853          * @param string $variant Language code of the variant to be checked for
 854          * @return bool True if $text appears to be written in $variant, false if not
 855          *
 856          * @author Nikola Smolenski <smolensk@eunet.rs>
 857          * @since 1.19
 858          */
 859         public function guessVariant( $text, $variant ) {
 860                 return false;
 861         }
 862
 863         /**
 864          * Load default conversion tables.
 865          * This method must be implemented in derived class.
 866          *
 867          * @private
 868          * @throws MWException
 869          */
 870         function loadDefaultTables() {
 871                 $class = static::class;
 872                 throw new MWException( "Must implement loadDefaultTables() method in class $class" );
 873         }
 874
 875         /**
 876          * Load conversion tables either from the cache or the disk.
 877          * @private
 878          * @param bool $fromCache Load from memcached? Defaults to true.
 879          */
 880         function loadTables( $fromCache = true ) {
 881                 global $wgLanguageConverterCacheType;
 882
 883                 if ( $this->mTablesLoaded ) {
 884                         return;
 885                 }
 886
 887                 $this->mTablesLoaded = true;
 888                 $this->mTables = false;
 889                 $cache = ObjectCache::getInstance( $wgLanguageConverterCacheType );
 890                 $cacheKey = $cache->makeKey( 'conversiontables', $this->mMainLanguageCode );
 891                 if ( $fromCache ) {
 892                         $this->mTables = $cache->get( $cacheKey );
 893                 }
 894                 if ( !$this->mTables || !array_key_exists( self::CACHE_VERSION_KEY, $this->mTables ) ) {
 895                         // not in cache, or we need a fresh reload.
 896                         // We will first load the default tables
 897                         // then update them using things in MediaWiki:Conversiontable/*
 898                         $this->loadDefaultTables();
 899                         foreach ( $this->mVariants as $var ) {
 900                                 $cached = $this->parseCachedTable( $var );
 901                                 $this->mTables[$var]->mergeArray( $cached );
 902                         }
 903
 904                         $this->postLoadTables();
 905                         $this->mTables[self::CACHE_VERSION_KEY] = true;
 906
 907                         $cache->set( $cacheKey, $this->mTables, 43200 );
 908                 }
 909         }
 910
 911         /**
 912          * Hook for post processing after conversion tables are loaded.
 913          */
 914         function postLoadTables() {
 915         }
 916
 917         /**
 918          * Reload the conversion tables.
 919          *
 920          * Also used by test suites which need to reset the converter state.
 921          *
 922          * @private
 923          */
 924         private function reloadTables() {
 925                 if ( $this->mTables ) {
 926                         unset( $this->mTables );
 927                 }
 928
 929                 $this->mTablesLoaded = false;
 930                 $this->loadTables( false );
 931         }
 932
 933         /**
 934          * Parse the conversion table stored in the cache.
 935          *
 936          * The tables should be in blocks of the following form:
 937          *              -{
 938          *                      word => word ;
 939          *                      word => word ;
 940          *                      ...
 941          *              }-
 942          *
 943          * To make the tables more manageable, subpages are allowed
 944          * and will be parsed recursively if $recursive == true.
 945          *
 946          * @param string $code Language code
 947          * @param string $subpage Subpage name
 948          * @param bool $recursive Parse subpages recursively? Defaults to true.
 949          *
 950          * @return array
 951          */
 952         function parseCachedTable( $code, $subpage = '', $recursive = true ) {
 953                 static $parsed = [];
 954
 955                 $key = 'Conversiontable/' . $code;
 956                 if ( $subpage ) {
 957                         $key .= '/' . $subpage;
 958                 }
 959                 if ( array_key_exists( $key, $parsed ) ) {
 960                         return [];
 961                 }
 962
 963                 $parsed[$key] = true;
 964
 965                 if ( $subpage === '' ) {
 966                         $txt = MessageCache::singleton()->getMsgFromNamespace( $key, $code );
 967                 } else {
 968                         $txt = false;
 969                         $title = Title::makeTitleSafe( NS_MEDIAWIKI, $key );
 970                         if ( $title && $title->exists() ) {
 971                                 $revision = Revision::newFromTitle( $title );
 972                                 if ( $revision ) {
 973                                         if ( $revision->getContentModel() == CONTENT_MODEL_WIKITEXT ) {
 974                                                 $txt = $revision->getContent( Revision::RAW )->getNativeData();
 975                                         }
 976
 977                                         // @todo in the future, use a specialized content model, perhaps based on json!
 978                                 }
 979                         }
 980                 }
 981
 982                 # Nothing to parse if there's no text
 983                 if ( $txt === false || $txt === null || $txt === '' ) {
 984                         return [];
 985                 }
 986
 987                 // get all subpage links of the form
 988                 // [[MediaWiki:Conversiontable/zh-xx/...|...]]
 989                 $linkhead = $this->mLangObj->getNsText( NS_MEDIAWIKI ) .
 990                         ':Conversiontable';
 991                 $subs = StringUtils::explode( '[[', $txt );
 992                 $sublinks = [];
 993                 foreach ( $subs as $sub ) {
 994                         $link = explode( ']]', $sub, 2 );
 995                         if ( count( $link ) != 2 ) {
 996                                 continue;
 997                         }
 998                         $b = explode( '|', $link[0], 2 );
 999                         $b = explode( '/', trim( $b[0] ), 3 );
1000                         if ( count( $b ) == 3 ) {
1001                                 $sublink = $b[2];
1002                         } else {
1003                                 $sublink = '';
1004                         }
1005
1006                         if ( $b[0] == $linkhead && $b[1] == $code ) {
1007                                 $sublinks[] = $sublink;
1008                         }
1009                 }
1010
1011                 // parse the mappings in this page
1012                 $blocks = StringUtils::explode( '-{', $txt );
1013                 $ret = [];
1014                 $first = true;
1015                 foreach ( $blocks as $block ) {
1016                         if ( $first ) {
1017                                 // Skip the part before the first -{
1018                                 $first = false;
1019                                 continue;
1020                         }
1021                         $mappings = explode( '}-', $block, 2 )[0];
1022                         $stripped = str_replace( [ "'", '"', '*', '#' ], '', $mappings );
1023                         $table = StringUtils::explode( ';', $stripped );
1024                         foreach ( $table as $t ) {
1025                                 $m = explode( '=>', $t, 3 );
1026                                 if ( count( $m ) != 2 ) {
1027                                         continue;
1028                                 }
1029                                 // trim any trailling comments starting with '//'
1030                                 $tt = explode( '//', $m[1], 2 );
1031                                 $ret[trim( $m[0] )] = trim( $tt[0] );
1032                         }
1033                 }
1034
1035                 // recursively parse the subpages
1036                 if ( $recursive ) {
1037                         foreach ( $sublinks as $link ) {
1038                                 $s = $this->parseCachedTable( $code, $link, $recursive );
1039                                 $ret = $s + $ret;
1040                         }
1041                 }
1042
1043                 if ( $this->mUcfirst ) {
1044                         foreach ( $ret as $k => $v ) {
1045                                 $ret[$this->mLangObj->ucfirst( $k )] = $this->mLangObj->ucfirst( $v );
1046                         }
1047                 }
1048                 return $ret;
1049         }
1050
1051         /**
1052          * Enclose a string with the "no conversion" tag. This is used by
1053          * various functions in the Parser.
1054          *
1055          * @param string $text Text to be tagged for no conversion
1056          * @param bool $noParse Unused
1057          * @return string The tagged text
1058          */
1059         public function markNoConversion( $text, $noParse = false ) {
1060                 # don't mark if already marked
1061                 if ( strpos( $text, '-{' ) || strpos( $text, '}-' ) ) {
1062                         return $text;
1063                 }
1064
1065                 $ret = "-{R|$text}-";
1066                 return $ret;
1067         }
1068
1069         /**
1070          * Convert the sorting key for category links. This should make different
1071          * keys that are variants of each other map to the same key.
1072          *
1073          * @param string $key
1074          *
1075          * @return string
1076          */
1077         function convertCategoryKey( $key ) {
1078                 return $key;
1079         }
1080
1081         /**
1082          * Refresh the cache of conversion tables when
1083          * MediaWiki:Conversiontable* is updated.
1084          *
1085          * @param Title $titleobj The Title of the page being updated
1086          */
1087         public function updateConversionTable( Title $titleobj ) {
1088                 if ( $titleobj->getNamespace() == NS_MEDIAWIKI ) {
1089                         $title = $titleobj->getDBkey();
1090                         $t = explode( '/', $title, 3 );
1091                         $c = count( $t );
1092                         if ( $c > 1 && $t[0] == 'Conversiontable' ) {
1093                                 if ( $this->validateVariant( $t[1] ) ) {
1094                                         $this->reloadTables();
1095                                 }
1096                         }
1097                 }
1098         }
1099
1100         /**
1101          * Get the cached separator pattern for ConverterRule::parseRules()
1102          * @return string
1103          */
1104         function getVarSeparatorPattern() {
1105                 if ( is_null( $this->mVarSeparatorPattern ) ) {
1106                         // varsep_pattern for preg_split:
1107                         // text should be splited by ";" only if a valid variant
1108                         // name exist after the markup, for example:
1109                         //  -{zh-hans:<span style="font-size:120%;">xxx</span>;zh-hant:\
1110                         //  <span style="font-size:120%;">yyy</span>;}-
1111                         // we should split it as:
1112                         //  [
1113                         //    [0] => 'zh-hans:<span style="font-size:120%;">xxx</span>'
1114                         //    [1] => 'zh-hant:<span style="font-size:120%;">yyy</span>'
1115                         //    [2] => ''
1116                         //  ]
1117                         $pat = '/;\s*(?=';
1118                         foreach ( $this->mVariants as $variant ) {
1119                                 // zh-hans:xxx;zh-hant:yyy
1120                                 $pat .= $variant . '\s*:|';
1121                                 // xxx=>zh-hans:yyy; xxx=>zh-hant:zzz
1122                                 $pat .= '[^;]*?=>\s*' . $variant . '\s*:|';
1123                         }
1124                         $pat .= '\s*$)/';
1125                         $this->mVarSeparatorPattern = $pat;
1126                 }
1127                 return $this->mVarSeparatorPattern;
1128         }
1129 }