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