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 use MediaWiki\Storage\RevisionRecord;
  25
  26 /**
  27  * Base class for language conversion.
  28  * @ingroup Language
  29  *
  30  * @author Zhengzhu Feng <zhengzhu@gmail.com>
  31  * @author fdcn <fdcn64@gmail.com>
  32  * @author shinjiman <shinjiman@gmail.com>
  33  * @author PhiLiP <philip.npc@gmail.com>
  34  */
  35 class LanguageConverter {
  36         /**
  37          * languages supporting variants
  38          * @since 1.20
  39          * @var array
  40          */
  41         public static $languagesWithVariants = [
  42                 'en',
  43                 'crh',
  44                 'gan',
  45                 'iu',
  46                 'kk',
  47                 'ku',
  48                 'shi',
  49                 'sr',
  50                 'tg',
  51                 'uz',
  52                 'zh',
  53         ];
  54
  55         public $mMainLanguageCode;
  56
  57         /**
  58          * @var string[]
  59          */
  60         public $mVariants;
  61         public $mVariantFallbacks;
  62         public $mVariantNames;
  63         public $mTablesLoaded = false;
  64
  65         /**
  66          * @var ReplacementArray[]
  67          * @phan-var array<string,ReplacementArray>
  68          */
  69         public $mTables;
  70
  71         // 'bidirectional' 'unidirectional' 'disable' for each variant
  72         public $mManualLevel;
  73
  74         public $mLangObj;
  75         public $mFlags;
  76         public $mDescCodeSep = ':', $mDescVarSep = ';';
  77         public $mUcfirst = false;
  78         public $mConvRuleTitle = false;
  79         public $mURLVariant;
  80         public $mUserVariant;
  81         public $mHeaderVariant;
  82         public $mMaxDepth = 10;
  83         public $mVarSeparatorPattern;
  84
  85         const CACHE_VERSION_KEY = 'VERSION 7';
  86
  87         /**
  88          * @param Language $langobj
  89          * @param string $maincode The main language code of this language
  90          * @param string[] $variants The supported variants of this language
  91          * @param array $variantfallbacks The fallback language of each variant
  92          * @param array $flags Defining the custom strings that maps to the flags
  93          * @param array $manualLevel Limit for supported variants
  94          */
  95         public function __construct( Language $langobj, $maincode, $variants = [],
  96                                                                 $variantfallbacks = [], $flags = [],
  97                                                                 $manualLevel = [] ) {
  98                 global $wgDisabledVariants;
  99                 $this->mLangObj = $langobj;
 100                 $this->mMainLanguageCode = $maincode;
 101                 $this->mVariants = array_diff( $variants, $wgDisabledVariants );
 102                 $this->mVariantFallbacks = $variantfallbacks;
 103                 $this->mVariantNames = Language::fetchLanguageNames();
 104                 $defaultflags = [
 105                         // 'S' show converted text
 106                         // '+' add rules for alltext
 107                         // 'E' the gave flags is error
 108                         // these flags above are reserved for program
 109                         'A' => 'A',   // add rule for convert code (all text convert)
 110                         'T' => 'T',   // title convert
 111                         'R' => 'R',   // raw content
 112                         'D' => 'D',   // convert description (subclass implement)
 113                         '-' => '-',   // remove convert (not implement)
 114                         'H' => 'H',   // add rule for convert code (but no display in placed code)
 115                         'N' => 'N',   // current variant name
 116                 ];
 117                 $this->mFlags = array_merge( $defaultflags, $flags );
 118                 foreach ( $this->mVariants as $v ) {
 119                         if ( array_key_exists( $v, $manualLevel ) ) {
 120                                 $this->mManualLevel[$v] = $manualLevel[$v];
 121                         } else {
 122                                 $this->mManualLevel[$v] = 'bidirectional';
 123                         }
 124                         $this->mFlags[$v] = $v;
 125                 }
 126         }
 127
 128         /**
 129          * Get all valid variants.
 130          * Call this instead of using $this->mVariants directly.
 131          *
 132          * @return string[] Contains all valid variants
 133          */
 134         public function getVariants() {
 135                 return $this->mVariants;
 136         }
 137
 138         /**
 139          * In case some variant is not defined in the markup, we need
 140          * to have some fallback. For example, in zh, normally people
 141          * will define zh-hans and zh-hant, but less so for zh-sg or zh-hk.
 142          * when zh-sg is preferred but not defined, we will pick zh-hans
 143          * in this case. Right now this is only used by zh.
 144          *
 145          * @param string $variant The language code of the variant
 146          * @return string|array The code of the fallback language or the
 147          *   main code if there is no fallback
 148          */
 149         public function getVariantFallbacks( $variant ) {
 150                 return $this->mVariantFallbacks[$variant] ?? $this->mMainLanguageCode;
 151         }
 152
 153         /**
 154          * Get the title produced by the conversion rule.
 155          * @return string The converted title text
 156          */
 157         public function getConvRuleTitle() {
 158                 return $this->mConvRuleTitle;
 159         }
 160
 161         /**
 162          * Get preferred language variant.
 163          * @return string The preferred language code
 164          */
 165         public function getPreferredVariant() {
 166                 global $wgDefaultLanguageVariant, $wgUser;
 167
 168                 $req = $this->getURLVariant();
 169
 170                 Hooks::run( 'GetLangPreferredVariant', [ &$req ] );
 171
 172                 if ( $wgUser->isSafeToLoad() && $wgUser->isLoggedIn() && !$req ) {
 173                         $req = $this->getUserVariant();
 174                 } elseif ( !$req ) {
 175                         $req = $this->getHeaderVariant();
 176                 }
 177
 178                 if ( $wgDefaultLanguageVariant && !$req ) {
 179                         $req = $this->validateVariant( $wgDefaultLanguageVariant );
 180                 }
 181
 182                 $req = $this->validateVariant( $req );
 183
 184                 // This function, unlike the other get*Variant functions, is
 185                 // not memoized (i.e. there return value is not cached) since
 186                 // new information might appear during processing after this
 187                 // is first called.
 188                 if ( $req ) {
 189                         return $req;
 190                 }
 191                 return $this->mMainLanguageCode;
 192         }
 193
 194         /**
 195          * Get default variant.
 196          * This function would not be affected by user's settings
 197          * @return string The default variant code
 198          */
 199         public function getDefaultVariant() {
 200                 global $wgDefaultLanguageVariant;
 201
 202                 $req = $this->getURLVariant();
 203
 204                 if ( !$req ) {
 205                         $req = $this->getHeaderVariant();
 206                 }
 207
 208                 if ( $wgDefaultLanguageVariant && !$req ) {
 209                         $req = $this->validateVariant( $wgDefaultLanguageVariant );
 210                 }
 211
 212                 if ( $req ) {
 213                         return $req;
 214                 }
 215                 return $this->mMainLanguageCode;
 216         }
 217
 218         /**
 219          * Validate the variant and return an appropriate strict internal
 220          * variant code if one exists.  Compare to Language::hasVariant()
 221          * which does a strict test.
 222          *
 223          * @param string|null $variant The variant to validate
 224          * @return mixed Returns an equivalent valid variant code if possible,
 225          *   null otherwise
 226          */
 227         public function validateVariant( $variant = null ) {
 228                 if ( $variant === null ) {
 229                         return null;
 230                 }
 231                 // Our internal variants are always lower-case; the variant we
 232                 // are validating may have mixed case.
 233                 $variant = LanguageCode::replaceDeprecatedCodes( strtolower( $variant ) );
 234                 if ( in_array( $variant, $this->mVariants ) ) {
 235                         return $variant;
 236                 }
 237                 // Browsers are supposed to use BCP 47 standard in the
 238                 // Accept-Language header, but not all of our internal
 239                 // mediawiki variant codes are BCP 47.  Map BCP 47 code
 240                 // to our internal code.
 241                 foreach ( $this->mVariants as $v ) {
 242                         // Case-insensitive match (BCP 47 is mixed case)
 243                         if ( strtolower( LanguageCode::bcp47( $v ) ) === $variant ) {
 244                                 return $v;
 245                         }
 246                 }
 247                 return null;
 248         }
 249
 250         /**
 251          * Get the variant specified in the URL
 252          *
 253          * @return mixed Variant if one found, null otherwise
 254          */
 255         public function getURLVariant() {
 256                 global $wgRequest;
 257
 258                 if ( $this->mURLVariant ) {
 259                         return $this->mURLVariant;
 260                 }
 261
 262                 // see if the preference is set in the request
 263                 $ret = $wgRequest->getText( 'variant' );
 264
 265                 if ( !$ret ) {
 266                         $ret = $wgRequest->getVal( 'uselang' );
 267                 }
 268
 269                 $this->mURLVariant = $this->validateVariant( $ret );
 270                 return $this->mURLVariant;
 271         }
 272
 273         /**
 274          * Determine if the user has a variant set.
 275          *
 276          * @return mixed Variant if one found, null otherwise
 277          */
 278         protected function getUserVariant() {
 279                 global $wgUser;
 280
 281                 // memoizing this function wreaks havoc on parserTest.php
 282                 /*
 283                 if ( $this->mUserVariant ) {
 284                         return $this->mUserVariant;
 285                 }
 286                 */
 287
 288                 // Get language variant preference from logged in users
 289                 // Don't call this on stub objects because that causes infinite
 290                 // recursion during initialisation
 291                 if ( !$wgUser->isSafeToLoad() ) {
 292                         return false;
 293                 }
 294                 if ( $wgUser->isLoggedIn() ) {
 295                         if (
 296                                 $this->mMainLanguageCode ==
 297                                 MediaWikiServices::getInstance()->getContentLanguage()->getCode()
 298                         ) {
 299                                 $ret = $wgUser->getOption( 'variant' );
 300                         } else {
 301                                 $ret = $wgUser->getOption( 'variant-' . $this->mMainLanguageCode );
 302                         }
 303                 } else {
 304                         // figure out user lang without constructing wgLang to avoid
 305                         // infinite recursion
 306                         $ret = $wgUser->getOption( 'language' );
 307                 }
 308
 309                 $this->mUserVariant = $this->validateVariant( $ret );
 310                 return $this->mUserVariant;
 311         }
 312
 313         /**
 314          * Determine the language variant from the Accept-Language header.
 315          *
 316          * @return mixed Variant if one found, null otherwise
 317          */
 318         protected function getHeaderVariant() {
 319                 global $wgRequest;
 320
 321                 if ( $this->mHeaderVariant ) {
 322                         return $this->mHeaderVariant;
 323                 }
 324
 325                 // See if some supported language variant is set in the
 326                 // HTTP header.
 327                 $languages = array_keys( $wgRequest->getAcceptLang() );
 328                 if ( empty( $languages ) ) {
 329                         return null;
 330                 }
 331
 332                 $fallbackLanguages = [];
 333                 foreach ( $languages as $language ) {
 334                         $this->mHeaderVariant = $this->validateVariant( $language );
 335                         if ( $this->mHeaderVariant ) {
 336                                 break;
 337                         }
 338
 339                         // To see if there are fallbacks of current language.
 340                         // We record these fallback variants, and process
 341                         // them later.
 342                         $fallbacks = $this->getVariantFallbacks( $language );
 343                         if ( is_string( $fallbacks ) && $fallbacks !== $this->mMainLanguageCode ) {
 344                                 $fallbackLanguages[] = $fallbacks;
 345                         } elseif ( is_array( $fallbacks ) ) {
 346                                 $fallbackLanguages =
 347                                         array_merge( $fallbackLanguages, $fallbacks );
 348                         }
 349                 }
 350
 351                 if ( !$this->mHeaderVariant ) {
 352                         // process fallback languages now
 353                         $fallback_languages = array_unique( $fallbackLanguages );
 354                         foreach ( $fallback_languages as $language ) {
 355                                 $this->mHeaderVariant = $this->validateVariant( $language );
 356                                 if ( $this->mHeaderVariant ) {
 357                                         break;
 358                                 }
 359                         }
 360                 }
 361
 362                 return $this->mHeaderVariant;
 363         }
 364
 365         /**
 366          * Dictionary-based conversion.
 367          * This function would not parse the conversion rules.
 368          * If you want to parse rules, try to use convert() or
 369          * convertTo().
 370          *
 371          * @param string $text The text to be converted
 372          * @param bool|string $toVariant The target language code
 373          * @return string The converted text
 374          */
 375         public function autoConvert( $text, $toVariant = false ) {
 376                 $this->loadTables();
 377
 378                 if ( !$toVariant ) {
 379                         $toVariant = $this->getPreferredVariant();
 380                         if ( !$toVariant ) {
 381                                 return $text;
 382                         }
 383                 }
 384
 385                 if ( $this->guessVariant( $text, $toVariant ) ) {
 386                         return $text;
 387                 }
 388                 /* we convert everything except:
 389                    1. HTML markups (anything between < and >)
 390                    2. HTML entities
 391                    3. placeholders created by the parser
 392                    IMPORTANT: Beware of failure from pcre.backtrack_limit (T124404).
 393                    Minimize use of backtracking where possible.
 394                 */
 395                 static $reg;
 396                 if ( $reg === null ) {
 397                         $marker = '|' . Parser::MARKER_PREFIX . '[^\x7f]++\x7f';
 398
 399                         // this one is needed when the text is inside an HTML markup
 400                         $htmlfix = '|<[^>\004]++(?=\004$)|^[^<>]*+>';
 401
 402                         // Optimize for the common case where these tags have
 403                         // few or no children. Thus try and possesively get as much as
 404                         // possible, and only engage in backtracking when we hit a '<'.
 405
 406                         // disable convert to variants between <code> tags
 407                         $codefix = '<code>[^<]*+(?:(?:(?!<\/code>).)[^<]*+)*+<\/code>|';
 408                         // disable conversion of <script> tags
 409                         $scriptfix = '<script[^>]*+>[^<]*+(?:(?:(?!<\/script>).)[^<]*+)*+<\/script>|';
 410                         // disable conversion of <pre> tags
 411                         $prefix = '<pre[^>]*+>[^<]*+(?:(?:(?!<\/pre>).)[^<]*+)*+<\/pre>|';
 412                         // The "|.*+)" at the end, is in case we missed some part of html syntax,
 413                         // we will fail securely (hopefully) by matching the rest of the string.
 414                         $htmlFullTag = '<(?:[^>=]*+(?>[^>=]*+=\s*+(?:"[^"]*"|\'[^\']*\'|[^\'">\s]*+))*+[^>=]*+>|.*+)|';
 415
 416                         $reg = '/' . $codefix . $scriptfix . $prefix . $htmlFullTag .
 417                                  '&[a-zA-Z#][a-z0-9]++;' . $marker . $htmlfix . '|\004$/s';
 418                 }
 419                 $startPos = 0;
 420                 $sourceBlob = '';
 421                 $literalBlob = '';
 422
 423                 // Guard against delimiter nulls in the input
 424                 // (should never happen: see T159174)
 425                 $text = str_replace( "\000", '', $text );
 426                 $text = str_replace( "\004", '', $text );
 427
 428                 $markupMatches = null;
 429                 $elementMatches = null;
 430
 431                 // We add a marker (\004) at the end of text, to ensure we always match the
 432                 // entire text (Otherwise, pcre.backtrack_limit might cause silent failure)
 433                 $textWithMarker = $text . "\004";
 434                 while ( $startPos < strlen( $text ) ) {
 435                         if ( preg_match( $reg, $textWithMarker, $markupMatches, PREG_OFFSET_CAPTURE, $startPos ) ) {
 436                                 $elementPos = $markupMatches[0][1];
 437                                 $element = $markupMatches[0][0];
 438                                 if ( $element === "\004" ) {
 439                                         // We hit the end.
 440                                         $elementPos = strlen( $text );
 441                                         $element = '';
 442                                 } elseif ( substr( $element, -1 ) === "\004" ) {
 443                                         // This can sometimes happen if we have
 444                                         // unclosed html tags (For example
 445                                         // when converting a title attribute
 446                                         // during a recursive call that contains
 447                                         // a &lt; e.g. <div title="&lt;">.
 448                                         $element = substr( $element, 0, -1 );
 449                                 }
 450                         } else {
 451                                 // If we hit here, then Language Converter could be tricked
 452                                 // into doing an XSS, so we refuse to translate.
 453                                 // If non-crazy input manages to reach this code path,
 454                                 // we should consider it a bug.
 455                                 $log = LoggerFactory::getInstance( 'languageconverter' );
 456                                 $log->error( "Hit pcre.backtrack_limit in " . __METHOD__
 457                                         . ". Disabling language conversion for this page.",
 458                                         [
 459                                                 "method" => __METHOD__,
 460                                                 "variant" => $toVariant,
 461                                                 "startOfText" => substr( $text, 0, 500 )
 462                                         ]
 463                                 );
 464                                 return $text;
 465                         }
 466                         // Queue the part before the markup for translation in a batch
 467                         $sourceBlob .= substr( $text, $startPos, $elementPos - $startPos ) . "\000";
 468
 469                         // Advance to the next position
 470                         $startPos = $elementPos + strlen( $element );
 471
 472                         // Translate any alt or title attributes inside the matched element
 473                         if ( $element !== ''
 474                                 && preg_match( '/^(<[^>\s]*+)\s([^>]*+)(.*+)$/', $element, $elementMatches )
 475                         ) {
 476                                 // FIXME, this decodes entities, so if you have something
 477                                 // like <div title="foo&lt;bar"> the bar won't get
 478                                 // translated since after entity decoding it looks like
 479                                 // unclosed html and we call this method recursively
 480                                 // on attributes.
 481                                 $attrs = Sanitizer::decodeTagAttributes( $elementMatches[2] );
 482                                 // Ensure self-closing tags stay self-closing.
 483                                 $close = substr( $elementMatches[2], -1 ) === '/' ? ' /' : '';
 484                                 $changed = false;
 485                                 foreach ( [ 'title', 'alt' ] as $attrName ) {
 486                                         if ( !isset( $attrs[$attrName] ) ) {
 487                                                 continue;
 488                                         }
 489                                         $attr = $attrs[$attrName];
 490                                         // Don't convert URLs
 491                                         if ( !strpos( $attr, '://' ) ) {
 492                                                 $attr = $this->recursiveConvertTopLevel( $attr, $toVariant );
 493                                         }
 494
 495                                         if ( $attr !== $attrs[$attrName] ) {
 496                                                 $attrs[$attrName] = $attr;
 497                                                 $changed = true;
 498                                         }
 499                                 }
 500                                 if ( $changed ) {
 501                                         $element = $elementMatches[1] . Html::expandAttributes( $attrs ) .
 502                                                 $close . $elementMatches[3];
 503                                 }
 504                         }
 505                         $literalBlob .= $element . "\000";
 506                 }
 507
 508                 // Do the main translation batch
 509                 $translatedBlob = $this->translate( $sourceBlob, $toVariant );
 510
 511                 // Put the output back together
 512                 $translatedIter = StringUtils::explode( "\000", $translatedBlob );
 513                 $literalIter = StringUtils::explode( "\000", $literalBlob );
 514                 $output = '';
 515                 while ( $translatedIter->valid() && $literalIter->valid() ) {
 516                         $output .= $translatedIter->current();
 517                         $output .= $literalIter->current();
 518                         $translatedIter->next();
 519                         $literalIter->next();
 520                 }
 521
 522                 return $output;
 523         }
 524
 525         /**
 526          * Translate a string to a variant.
 527          * Doesn't parse rules or do any of that other stuff, for that use
 528          * convert() or convertTo().
 529          *
 530          * @param string $text Text to convert
 531          * @param string $variant Variant language code
 532          * @return string Translated text
 533          */
 534         public function translate( $text, $variant ) {
 535                 // If $text is empty or only includes spaces, do nothing
 536                 // Otherwise translate it
 537                 if ( trim( $text ) ) {
 538                         $this->loadTables();
 539                         $text = $this->mTables[$variant]->replace( $text );
 540                 }
 541                 return $text;
 542         }
 543
 544         /**
 545          * Call translate() to convert text to all valid variants.
 546          *
 547          * @param string $text The text to be converted
 548          * @return array Variant => converted text
 549          */
 550         public function autoConvertToAllVariants( $text ) {
 551                 $this->loadTables();
 552
 553                 $ret = [];
 554                 foreach ( $this->mVariants as $variant ) {
 555                         $ret[$variant] = $this->translate( $text, $variant );
 556                 }
 557
 558                 return $ret;
 559         }
 560
 561         /**
 562          * Apply manual conversion rules.
 563          *
 564          * @param ConverterRule $convRule
 565          */
 566         protected function applyManualConv( $convRule ) {
 567                 // Use syntax -{T|zh-cn:TitleCN; zh-tw:TitleTw}- to custom
 568                 // title conversion.
 569                 // T26072: $mConvRuleTitle was overwritten by other manual
 570                 // rule(s) not for title, this breaks the title conversion.
 571                 $newConvRuleTitle = $convRule->getTitle();
 572                 if ( $newConvRuleTitle ) {
 573                         // So I add an empty check for getTitle()
 574                         $this->mConvRuleTitle = $newConvRuleTitle;
 575                 }
 576
 577                 // merge/remove manual conversion rules to/from global table
 578                 $convTable = $convRule->getConvTable();
 579                 $action = $convRule->getRulesAction();
 580                 foreach ( $convTable as $variant => $pair ) {
 581                         $v = $this->validateVariant( $variant );
 582                         if ( !$v ) {
 583                                 continue;
 584                         }
 585
 586                         if ( $action == 'add' ) {
 587                                 // More efficient than array_merge(), about 2.5 times.
 588                                 foreach ( $pair as $from => $to ) {
 589                                         $this->mTables[$v]->setPair( $from, $to );
 590                                 }
 591                         } elseif ( $action == 'remove' ) {
 592                                 $this->mTables[$v]->removeArray( $pair );
 593                         }
 594                 }
 595         }
 596
 597         /**
 598          * Auto convert a Title object to a readable string in the
 599          * preferred variant.
 600          *
 601          * @param Title $title A object of Title
 602          * @return string Converted title text
 603          */
 604         public function convertTitle( $title ) {
 605                 $variant = $this->getPreferredVariant();
 606                 $index = $title->getNamespace();
 607                 if ( $index !== NS_MAIN ) {
 608                         $text = $this->convertNamespace( $index, $variant ) . ':';
 609                 } else {
 610                         $text = '';
 611                 }
 612                 $text .= $this->translate( $title->getText(), $variant );
 613                 return $text;
 614         }
 615
 616         /**
 617          * Get the namespace display name in the preferred variant.
 618          *
 619          * @param int $index Namespace id
 620          * @param string|null $variant Variant code or null for preferred variant
 621          * @return string Namespace name for display
 622          */
 623         public function convertNamespace( $index, $variant = null ) {
 624                 if ( $index === NS_MAIN ) {
 625                         return '';
 626                 }
 627
 628                 if ( $variant === null ) {
 629                         $variant = $this->getPreferredVariant();
 630                 }
 631
 632                 $cache = MediaWikiServices::getInstance()->getLocalServerObjectCache();
 633                 $key = $cache->makeKey( 'languageconverter', 'namespace-text', $index, $variant );
 634                 $nsVariantText = $cache->get( $key );
 635                 if ( $nsVariantText !== false ) {
 636                         return $nsVariantText;
 637                 }
 638
 639                 // First check if a message gives a converted name in the target variant.
 640                 $nsConvMsg = wfMessage( 'conversion-ns' . $index )->inLanguage( $variant );
 641                 if ( $nsConvMsg->exists() ) {
 642                         $nsVariantText = $nsConvMsg->plain();
 643                 }
 644
 645                 // Then check if a message gives a converted name in content language
 646                 // which needs extra translation to the target variant.
 647                 if ( $nsVariantText === false ) {
 648                         $nsConvMsg = wfMessage( 'conversion-ns' . $index )->inContentLanguage();
 649                         if ( $nsConvMsg->exists() ) {
 650                                 $nsVariantText = $this->translate( $nsConvMsg->plain(), $variant );
 651                         }
 652                 }
 653
 654                 if ( $nsVariantText === false ) {
 655                         // No message exists, retrieve it from the target variant's namespace names.
 656                         $langObj = $this->mLangObj->factory( $variant );
 657                         $nsVariantText = $langObj->getFormattedNsText( $index );
 658                 }
 659
 660                 $cache->set( $key, $nsVariantText, 60 );
 661
 662                 return $nsVariantText;
 663         }
 664
 665         /**
 666          * Convert text to different variants of a language. The automatic
 667          * conversion is done in autoConvert(). Here we parse the text
 668          * marked with -{}-, which specifies special conversions of the
 669          * text that can not be accomplished in autoConvert().
 670          *
 671          * Syntax of the markup:
 672          * -{code1:text1;code2:text2;...}-  or
 673          * -{flags|code1:text1;code2:text2;...}-  or
 674          * -{text}- in which case no conversion should take place for text
 675          *
 676          * @warning Glossary state is maintained between calls. Never feed this
 677          *   method input that hasn't properly been escaped as it may result in
 678          *   an XSS in subsequent calls, even if those subsequent calls properly
 679          *   escape things.
 680          * @param string $text Text to be converted, already html escaped.
 681          * @return string Converted text (html)
 682          */
 683         public function convert( $text ) {
 684                 $variant = $this->getPreferredVariant();
 685                 return $this->convertTo( $text, $variant );
 686         }
 687
 688         /**
 689          * Same as convert() except a extra parameter to custom variant.
 690          *
 691          * @param string $text Text to be converted, already html escaped
 692          * @param-taint $text exec_html
 693          * @param string $variant The target variant code
 694          * @return string Converted text
 695          * @return-taint escaped
 696          */
 697         public function convertTo( $text, $variant ) {
 698                 global $wgDisableLangConversion;
 699                 if ( $wgDisableLangConversion ) {
 700                         return $text;
 701                 }
 702                 // Reset converter state for a new converter run.
 703                 $this->mConvRuleTitle = false;
 704                 return $this->recursiveConvertTopLevel( $text, $variant );
 705         }
 706
 707         /**
 708          * Recursively convert text on the outside. Allow to use nested
 709          * markups to custom rules.
 710          *
 711          * @param string $text Text to be converted
 712          * @param string $variant The target variant code
 713          * @param int $depth Depth of recursion
 714          * @return string Converted text
 715          */
 716         protected function recursiveConvertTopLevel( $text, $variant, $depth = 0 ) {
 717                 $startPos = 0;
 718                 $out = '';
 719                 $length = strlen( $text );
 720                 $shouldConvert = !$this->guessVariant( $text, $variant );
 721                 $continue = 1;
 722
 723                 $noScript = '<script.*?>.*?<\/script>(*SKIP)(*FAIL)';
 724                 $noStyle = '<style.*?>.*?<\/style>(*SKIP)(*FAIL)';
 725                 // phpcs:ignore Generic.Files.LineLength
 726                 $noHtml = '<(?:[^>=]*+(?>[^>=]*+=\s*+(?:"[^"]*"|\'[^\']*\'|[^\'">\s]*+))*+[^>=]*+>|.*+)(*SKIP)(*FAIL)';
 727                 while ( $startPos < $length && $continue ) {
 728                         $continue = preg_match(
 729                                 // Only match -{ outside of html.
 730                                 "/$noScript|$noStyle|$noHtml|-\{/",
 731                                 $text,
 732                                 $m,
 733                                 PREG_OFFSET_CAPTURE,
 734                                 $startPos
 735                         );
 736
 737                         if ( !$continue ) {
 738                                 // No more markup, append final segment
 739                                 $fragment = substr( $text, $startPos );
 740                                 $out .= $shouldConvert ? $this->autoConvert( $fragment, $variant ) : $fragment;
 741                                 return $out;
 742                         }
 743
 744                         // Offset of the match of the regex pattern.
 745                         $pos = $m[0][1];
 746
 747                         // Append initial segment
 748                         $fragment = substr( $text, $startPos, $pos - $startPos );
 749                         $out .= $shouldConvert ? $this->autoConvert( $fragment, $variant ) : $fragment;
 750                         // -{ marker found, not in attribute
 751                         // Advance position up to -{ marker.
 752                         $startPos = $pos;
 753                         // Do recursive conversion
 754                         // Note: This passes $startPos by reference, and advances it.
 755                         $out .= $this->recursiveConvertRule( $text, $variant, $startPos, $depth + 1 );
 756                 }
 757                 return $out;
 758         }
 759
 760         /**
 761          * Recursively convert text on the inside.
 762          *
 763          * @param string $text Text to be converted
 764          * @param string $variant The target variant code
 765          * @param int &$startPos
 766          * @param int $depth Depth of recursion
 767          *
 768          * @throws MWException
 769          * @return string Converted text
 770          */
 771         protected function recursiveConvertRule( $text, $variant, &$startPos, $depth = 0 ) {
 772                 // Quick sanity check (no function calls)
 773                 if ( $text[$startPos] !== '-' || $text[$startPos + 1] !== '{' ) {
 774                         throw new MWException( __METHOD__ . ': invalid input string' );
 775                 }
 776
 777                 $startPos += 2;
 778                 $inner = '';
 779                 $warningDone = false;
 780                 $length = strlen( $text );
 781
 782                 while ( $startPos < $length ) {
 783                         $m = false;
 784                         preg_match( '/-\{|\}-/', $text, $m, PREG_OFFSET_CAPTURE, $startPos );
 785                         if ( !$m ) {
 786                                 // Unclosed rule
 787                                 break;
 788                         }
 789
 790                         $token = $m[0][0];
 791                         $pos = $m[0][1];
 792
 793                         // Markup found
 794                         // Append initial segment
 795                         $inner .= substr( $text, $startPos, $pos - $startPos );
 796
 797                         // Advance position
 798                         $startPos = $pos;
 799
 800                         switch ( $token ) {
 801                                 case '-{':
 802                                         // Check max depth
 803                                         if ( $depth >= $this->mMaxDepth ) {
 804                                                 $inner .= '-{';
 805                                                 if ( !$warningDone ) {
 806                                                         $inner .= '<span class="error">' .
 807                                                                 wfMessage( 'language-converter-depth-warning' )
 808                                                                         ->numParams( $this->mMaxDepth )->inContentLanguage()->text() .
 809                                                                 '</span>';
 810                                                         $warningDone = true;
 811                                                 }
 812                                                 $startPos += 2;
 813                                                 break;
 814                                         }
 815                                         // Recursively parse another rule
 816                                         $inner .= $this->recursiveConvertRule( $text, $variant, $startPos, $depth + 1 );
 817                                         break;
 818                                 case '}-':
 819                                         // Apply the rule
 820                                         $startPos += 2;
 821                                         $rule = new ConverterRule( $inner, $this );
 822                                         $rule->parse( $variant );
 823                                         $this->applyManualConv( $rule );
 824                                         return $rule->getDisplay();
 825                                 default:
 826                                         throw new MWException( __METHOD__ . ': invalid regex match' );
 827                         }
 828                 }
 829
 830                 // Unclosed rule
 831                 if ( $startPos < $length ) {
 832                         $inner .= substr( $text, $startPos );
 833                 }
 834                 $startPos = $length;
 835                 return '-{' . $this->autoConvert( $inner, $variant );
 836         }
 837
 838         /**
 839          * If a language supports multiple variants, it is possible that
 840          * non-existing link in one variant actually exists in another variant.
 841          * This function tries to find it. See e.g. LanguageZh.php
 842          * The input parameters may be modified upon return
 843          *
 844          * @param string &$link The name of the link
 845          * @param Title &$nt The title object of the link
 846          * @param bool $ignoreOtherCond To disable other conditions when
 847          *   we need to transclude a template or update a category's link
 848          */
 849         public function findVariantLink( &$link, &$nt, $ignoreOtherCond = false ) {
 850                 # If the article has already existed, there is no need to
 851                 # check it again, otherwise it may cause a fault.
 852                 if ( is_object( $nt ) && $nt->exists() ) {
 853                         return;
 854                 }
 855
 856                 global $wgDisableLangConversion, $wgDisableTitleConversion, $wgRequest;
 857                 $isredir = $wgRequest->getText( 'redirect', 'yes' );
 858                 $action = $wgRequest->getText( 'action' );
 859                 if ( $action == 'edit' && $wgRequest->getBool( 'redlink' ) ) {
 860                         $action = 'view';
 861                 }
 862                 $linkconvert = $wgRequest->getText( 'linkconvert', 'yes' );
 863                 $disableLinkConversion = $wgDisableLangConversion
 864                         || $wgDisableTitleConversion;
 865                 $linkBatch = new LinkBatch();
 866
 867                 $ns = NS_MAIN;
 868
 869                 if ( $disableLinkConversion ||
 870                         ( !$ignoreOtherCond &&
 871                                 ( $isredir == 'no'
 872                                         || $action == 'edit'
 873                                         || $action == 'submit'
 874                                         || $linkconvert == 'no' ) ) ) {
 875                         return;
 876                 }
 877
 878                 if ( is_object( $nt ) ) {
 879                         $ns = $nt->getNamespace();
 880                 }
 881
 882                 $variants = $this->autoConvertToAllVariants( $link );
 883                 if ( !$variants ) { // give up
 884                         return;
 885                 }
 886
 887                 $titles = [];
 888
 889                 foreach ( $variants as $v ) {
 890                         if ( $v != $link ) {
 891                                 $varnt = Title::newFromText( $v, $ns );
 892                                 if ( !is_null( $varnt ) ) {
 893                                         $linkBatch->addObj( $varnt );
 894                                         $titles[] = $varnt;
 895                                 }
 896                         }
 897                 }
 898
 899                 // fetch all variants in single query
 900                 $linkBatch->execute();
 901
 902                 foreach ( $titles as $varnt ) {
 903                         if ( $varnt->getArticleID() > 0 ) {
 904                                 $nt = $varnt;
 905                                 $link = $varnt->getText();
 906                                 break;
 907                         }
 908                 }
 909         }
 910
 911         /**
 912          * Returns language specific hash options.
 913          *
 914          * @return string
 915          */
 916         public function getExtraHashOptions() {
 917                 $variant = $this->getPreferredVariant();
 918
 919                 return '!' . $variant;
 920         }
 921
 922         /**
 923          * Guess if a text is written in a variant. This should be implemented in subclasses.
 924          *
 925          * @param string $text The text to be checked
 926          * @param string $variant Language code of the variant to be checked for
 927          * @return bool True if $text appears to be written in $variant, false if not
 928          *
 929          * @author Nikola Smolenski <smolensk@eunet.rs>
 930          * @since 1.19
 931          */
 932         public function guessVariant( $text, $variant ) {
 933                 return false;
 934         }
 935
 936         /**
 937          * Load default conversion tables.
 938          * This method must be implemented in derived class.
 939          *
 940          * @private
 941          * @throws MWException
 942          */
 943         function loadDefaultTables() {
 944                 $class = static::class;
 945                 throw new MWException( "Must implement loadDefaultTables() method in class $class" );
 946         }
 947
 948         /**
 949          * Load conversion tables either from the cache or the disk.
 950          * @private
 951          * @param bool $fromCache Load from memcached? Defaults to true.
 952          */
 953         function loadTables( $fromCache = true ) {
 954                 global $wgLanguageConverterCacheType;
 955
 956                 if ( $this->mTablesLoaded ) {
 957                         return;
 958                 }
 959
 960                 $this->mTablesLoaded = true;
 961                 $this->mTables = false;
 962                 $cache = ObjectCache::getInstance( $wgLanguageConverterCacheType );
 963                 $cacheKey = $cache->makeKey( 'conversiontables', $this->mMainLanguageCode );
 964                 if ( $fromCache ) {
 965                         $this->mTables = $cache->get( $cacheKey );
 966                 }
 967                 if ( !$this->mTables || !array_key_exists( self::CACHE_VERSION_KEY, $this->mTables ) ) {
 968                         // not in cache, or we need a fresh reload.
 969                         // We will first load the default tables
 970                         // then update them using things in MediaWiki:Conversiontable/*
 971                         $this->loadDefaultTables();
 972                         foreach ( $this->mVariants as $var ) {
 973                                 $cached = $this->parseCachedTable( $var );
 974                                 $this->mTables[$var]->mergeArray( $cached );
 975                         }
 976
 977                         $this->postLoadTables();
 978                         $this->mTables[self::CACHE_VERSION_KEY] = true;
 979
 980                         $cache->set( $cacheKey, $this->mTables, 43200 );
 981                 }
 982         }
 983
 984         /**
 985          * Hook for post processing after conversion tables are loaded.
 986          */
 987         function postLoadTables() {
 988         }
 989
 990         /**
 991          * Reload the conversion tables.
 992          *
 993          * Also used by test suites which need to reset the converter state.
 994          *
 995          * @private
 996          */
 997         private function reloadTables() {
 998                 if ( $this->mTables ) {
 999                         // @phan-suppress-next-line PhanTypeObjectUnsetDeclaredProperty
1000                         unset( $this->mTables );
1001                 }
1002
1003                 $this->mTablesLoaded = false;
1004                 $this->loadTables( false );
1005         }
1006
1007         /**
1008          * Parse the conversion table stored in the cache.
1009          *
1010          * The tables should be in blocks of the following form:
1011          *              -{
1012          *                      word => word ;
1013          *                      word => word ;
1014          *                      ...
1015          *              }-
1016          *
1017          * To make the tables more manageable, subpages are allowed
1018          * and will be parsed recursively if $recursive == true.
1019          *
1020          * @param string $code Language code
1021          * @param string $subpage Subpage name
1022          * @param bool $recursive Parse subpages recursively? Defaults to true.
1023          *
1024          * @return array
1025          */
1026         function parseCachedTable( $code, $subpage = '', $recursive = true ) {
1027                 static $parsed = [];
1028
1029                 $key = 'Conversiontable/' . $code;
1030                 if ( $subpage ) {
1031                         $key .= '/' . $subpage;
1032                 }
1033                 if ( array_key_exists( $key, $parsed ) ) {
1034                         return [];
1035                 }
1036
1037                 $parsed[$key] = true;
1038
1039                 if ( $subpage === '' ) {
1040                         $txt = MessageCache::singleton()->getMsgFromNamespace( $key, $code );
1041                 } else {
1042                         $txt = false;
1043                         $title = Title::makeTitleSafe( NS_MEDIAWIKI, $key );
1044                         if ( $title && $title->exists() ) {
1045                                 $revision = Revision::newFromTitle( $title );
1046                                 if ( $revision ) {
1047                                         if ( $revision->getContentModel() == CONTENT_MODEL_WIKITEXT ) {
1048                                                 $txt = $revision->getContent( RevisionRecord::RAW )->getText();
1049                                         }
1050
1051                                         // @todo in the future, use a specialized content model, perhaps based on json!
1052                                 }
1053                         }
1054                 }
1055
1056                 # Nothing to parse if there's no text
1057                 if ( $txt === false || $txt === null || $txt === '' ) {
1058                         return [];
1059                 }
1060
1061                 // get all subpage links of the form
1062                 // [[MediaWiki:Conversiontable/zh-xx/...|...]]
1063                 $linkhead = $this->mLangObj->getNsText( NS_MEDIAWIKI ) .
1064                         ':Conversiontable';
1065                 $subs = StringUtils::explode( '[[', $txt );
1066                 $sublinks = [];
1067                 foreach ( $subs as $sub ) {
1068                         $link = explode( ']]', $sub, 2 );
1069                         if ( count( $link ) != 2 ) {
1070                                 continue;
1071                         }
1072                         $b = explode( '|', $link[0], 2 );
1073                         $b = explode( '/', trim( $b[0] ), 3 );
1074                         if ( count( $b ) == 3 ) {
1075                                 $sublink = $b[2];
1076                         } else {
1077                                 $sublink = '';
1078                         }
1079
1080                         if ( $b[0] == $linkhead && $b[1] == $code ) {
1081                                 $sublinks[] = $sublink;
1082                         }
1083                 }
1084
1085                 // parse the mappings in this page
1086                 $blocks = StringUtils::explode( '-{', $txt );
1087                 $ret = [];
1088                 $first = true;
1089                 foreach ( $blocks as $block ) {
1090                         if ( $first ) {
1091                                 // Skip the part before the first -{
1092                                 $first = false;
1093                                 continue;
1094                         }
1095                         $mappings = explode( '}-', $block, 2 )[0];
1096                         $stripped = str_replace( [ "'", '"', '*', '#' ], '', $mappings );
1097                         $table = StringUtils::explode( ';', $stripped );
1098                         foreach ( $table as $t ) {
1099                                 $m = explode( '=>', $t, 3 );
1100                                 if ( count( $m ) != 2 ) {
1101                                         continue;
1102                                 }
1103                                 // trim any trailling comments starting with '//'
1104                                 $tt = explode( '//', $m[1], 2 );
1105                                 $ret[trim( $m[0] )] = trim( $tt[0] );
1106                         }
1107                 }
1108
1109                 // recursively parse the subpages
1110                 if ( $recursive ) {
1111                         foreach ( $sublinks as $link ) {
1112                                 $s = $this->parseCachedTable( $code, $link, $recursive );
1113                                 $ret = $s + $ret;
1114                         }
1115                 }
1116
1117                 if ( $this->mUcfirst ) {
1118                         foreach ( $ret as $k => $v ) {
1119                                 $ret[$this->mLangObj->ucfirst( $k )] = $this->mLangObj->ucfirst( $v );
1120                         }
1121                 }
1122                 return $ret;
1123         }
1124
1125         /**
1126          * Enclose a string with the "no conversion" tag. This is used by
1127          * various functions in the Parser.
1128          *
1129          * @param string $text Text to be tagged for no conversion
1130          * @param bool $noParse Unused
1131          * @return string The tagged text
1132          */
1133         public function markNoConversion( $text, $noParse = false ) {
1134                 # don't mark if already marked
1135                 if ( strpos( $text, '-{' ) || strpos( $text, '}-' ) ) {
1136                         return $text;
1137                 }
1138
1139                 $ret = "-{R|$text}-";
1140                 return $ret;
1141         }
1142
1143         /**
1144          * Convert the sorting key for category links. This should make different
1145          * keys that are variants of each other map to the same key.
1146          *
1147          * @param string $key
1148          *
1149          * @return string
1150          */
1151         function convertCategoryKey( $key ) {
1152                 return $key;
1153         }
1154
1155         /**
1156          * Refresh the cache of conversion tables when
1157          * MediaWiki:Conversiontable* is updated.
1158          *
1159          * @param Title $titleobj The Title of the page being updated
1160          */
1161         public function updateConversionTable( Title $titleobj ) {
1162                 if ( $titleobj->getNamespace() == NS_MEDIAWIKI ) {
1163                         $title = $titleobj->getDBkey();
1164                         $t = explode( '/', $title, 3 );
1165                         $c = count( $t );
1166                         if ( $c > 1 && $t[0] == 'Conversiontable' ) {
1167                                 if ( $this->validateVariant( $t[1] ) ) {
1168                                         $this->reloadTables();
1169                                 }
1170                         }
1171                 }
1172         }
1173
1174         /**
1175          * Get the cached separator pattern for ConverterRule::parseRules()
1176          * @return string
1177          */
1178         function getVarSeparatorPattern() {
1179                 if ( is_null( $this->mVarSeparatorPattern ) ) {
1180                         // varsep_pattern for preg_split:
1181                         // text should be splited by ";" only if a valid variant
1182                         // name exist after the markup, for example:
1183                         //  -{zh-hans:<span style="font-size:120%;">xxx</span>;zh-hant:\
1184                         //  <span style="font-size:120%;">yyy</span>;}-
1185                         // we should split it as:
1186                         //  [
1187                         //    [0] => 'zh-hans:<span style="font-size:120%;">xxx</span>'
1188                         //    [1] => 'zh-hant:<span style="font-size:120%;">yyy</span>'
1189                         //    [2] => ''
1190                         //  ]
1191                         $expandedVariants = [];
1192                         foreach ( $this->mVariants as $variant ) {
1193                                 $expandedVariants[ $variant ] = 1;
1194                                 // Accept standard BCP 47 names for variants as well.
1195                                 $expandedVariants[ LanguageCode::bcp47( $variant ) ] = 1;
1196                         }
1197                         // Accept old deprecated names for variants
1198                         foreach ( LanguageCode::getDeprecatedCodeMapping() as $old => $new ) {
1199                                 if ( isset( $expandedVariants[ $new ] ) ) {
1200                                         $expandedVariants[ $old ] = 1;
1201                                 }
1202                         }
1203
1204                         $pat = '/;\s*(?=';
1205                         foreach ( $expandedVariants as $variant => $ignore ) {
1206                                 // zh-hans:xxx;zh-hant:yyy
1207                                 $pat .= $variant . '\s*:|';
1208                                 // xxx=>zh-hans:yyy; xxx=>zh-hant:zzz
1209                                 $pat .= '[^;]*?=>\s*' . $variant . '\s*:|';
1210                         }
1211                         $pat .= '\s*$)/';
1212                         $this->mVarSeparatorPattern = $pat;
1213                 }
1214                 return $this->mVarSeparatorPattern;
1215         }
1216 }