languages/LanguageConverter.php

   1 <?php
   2 /**
   3  * This program is free software; you can redistribute it and/or modify
   4  * it under the terms of the GNU General Public License as published by
   5  * the Free Software Foundation; either version 2 of the License, or
   6  * (at your option) any later version.
   7  *
   8  * This program is distributed in the hope that it will be useful,
   9  * but WITHOUT ANY WARRANTY; without even the implied warranty of
  10  * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the
  11  * GNU General Public License for more details.
  12  *
  13  * You should have received a copy of the GNU General Public License along
  14  * with this program; if not, write to the Free Software Foundation, Inc.,
  15  * 51 Franklin Street, Fifth Floor, Boston, MA 02110-1301, USA.
  16  * http://www.gnu.org/copyleft/gpl.html
  17  *
  18  * @file
  19  * @ingroup Language
  20  */
  21 use MediaWiki\MediaWikiServices;
  22
  23 use MediaWiki\Logger\LoggerFactory;
  24
  25 /**
  26  * Base class for language conversion.
  27  * @ingroup Language
  28  *
  29  * @author Zhengzhu Feng <zhengzhu@gmail.com>
  30  * @author fdcn <fdcn64@gmail.com>
  31  * @author shinjiman <shinjiman@gmail.com>
  32  * @author PhiLiP <philip.npc@gmail.com>
  33  */
  34 class LanguageConverter {
  35         /**
  36          * languages supporting variants
  37          * @since 1.20
  38          * @var array
  39          */
  40         static public $languagesWithVariants = [
  41                 'en',
  42                 '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         public $mTables;
  64         // 'bidirectional' 'unidirectional' 'disable' for each variant
  65         public $mManualLevel;
  66
  67         public $mLangObj;
  68         public $mFlags;
  69         public $mDescCodeSep = ':', $mDescVarSep = ';';
  70         public $mUcfirst = false;
  71         public $mConvRuleTitle = false;
  72         public $mURLVariant;
  73         public $mUserVariant;
  74         public $mHeaderVariant;
  75         public $mMaxDepth = 10;
  76         public $mVarSeparatorPattern;
  77
  78         const CACHE_VERSION_KEY = 'VERSION 7';
  79
  80         /**
  81          * @param Language $langobj
  82          * @param string $maincode The main language code of this language
  83          * @param string[] $variants The supported variants of this language
  84          * @param array $variantfallbacks The fallback language of each variant
  85          * @param array $flags Defining the custom strings that maps to the flags
  86          * @param array $manualLevel Limit for supported variants
  87          */
  88         public function __construct( Language $langobj, $maincode, $variants = [],
  89                                                                 $variantfallbacks = [], $flags = [],
  90                                                                 $manualLevel = [] ) {
  91                 global $wgDisabledVariants;
  92                 $this->mLangObj = $langobj;
  93                 $this->mMainLanguageCode = $maincode;
  94                 $this->mVariants = array_diff( $variants, $wgDisabledVariants );
  95                 $this->mVariantFallbacks = $variantfallbacks;
  96                 $this->mVariantNames = Language::fetchLanguageNames();
  97                 $defaultflags = [
  98                         // 'S' show converted text
  99                         // '+' add rules for alltext
 100                         // 'E' the gave flags is error
 101                         // these flags above are reserved for program
 102                         'A' => 'A',   // add rule for convert code (all text convert)
 103                         'T' => 'T',   // title convert
 104                         'R' => 'R',   // raw content
 105                         'D' => 'D',   // convert description (subclass implement)
 106                         '-' => '-',   // remove convert (not implement)
 107                         'H' => 'H',   // add rule for convert code (but no display in placed code)
 108                         'N' => 'N',   // current variant name
 109                 ];
 110                 $this->mFlags = array_merge( $defaultflags, $flags );
 111                 foreach ( $this->mVariants as $v ) {
 112                         if ( array_key_exists( $v, $manualLevel ) ) {
 113                                 $this->mManualLevel[$v] = $manualLevel[$v];
 114                         } else {
 115                                 $this->mManualLevel[$v] = 'bidirectional';
 116                         }
 117                         $this->mFlags[$v] = $v;
 118                 }
 119         }
 120
 121         /**
 122          * Get all valid variants.
 123          * Call this instead of using $this->mVariants directly.
 124          *
 125          * @return string[] Contains all valid variants
 126          */
 127         public function getVariants() {
 128                 return $this->mVariants;
 129         }
 130
 131         /**
 132          * In case some variant is not defined in the markup, we need
 133          * to have some fallback. For example, in zh, normally people
 134          * will define zh-hans and zh-hant, but less so for zh-sg or zh-hk.
 135          * when zh-sg is preferred but not defined, we will pick zh-hans
 136          * in this case. Right now this is only used by zh.
 137          *
 138          * @param string $variant The language code of the variant
 139          * @return string|array The code of the fallback language or the
 140          *   main code if there is no fallback
 141          */
 142         public function getVariantFallbacks( $variant ) {
 143                 if ( isset( $this->mVariantFallbacks[$variant] ) ) {
 144                         return $this->mVariantFallbacks[$variant];
 145                 }
 146                 return $this->mMainLanguageCode;
 147         }
 148
 149         /**
 150          * Get the title produced by the conversion rule.
 151          * @return string The converted title text
 152          */
 153         public function getConvRuleTitle() {
 154                 return $this->mConvRuleTitle;
 155         }
 156
 157         /**
 158          * Get preferred language variant.
 159          * @return string The preferred language code
 160          */
 161         public function getPreferredVariant() {
 162                 global $wgDefaultLanguageVariant, $wgUser;
 163
 164                 $req = $this->getURLVariant();
 165
 166                 Hooks::run( 'GetLangPreferredVariant', [ &$req ] );
 167
 168                 if ( $wgUser->isSafeToLoad() && $wgUser->isLoggedIn() && !$req ) {
 169                         $req = $this->getUserVariant();
 170                 } elseif ( !$req ) {
 171                         $req = $this->getHeaderVariant();
 172                 }
 173
 174                 if ( $wgDefaultLanguageVariant && !$req ) {
 175                         $req = $this->validateVariant( $wgDefaultLanguageVariant );
 176                 }
 177
 178                 $req = $this->validateVariant( $req );
 179
 180                 // This function, unlike the other get*Variant functions, is
 181                 // not memoized (i.e. there return value is not cached) since
 182                 // new information might appear during processing after this
 183                 // is first called.
 184                 if ( $req ) {
 185                         return $req;
 186                 }
 187                 return $this->mMainLanguageCode;
 188         }
 189
 190         /**
 191          * Get default variant.
 192          * This function would not be affected by user's settings
 193          * @return string The default variant code
 194          */
 195         public function getDefaultVariant() {
 196                 global $wgDefaultLanguageVariant;
 197
 198                 $req = $this->getURLVariant();
 199
 200                 if ( !$req ) {
 201                         $req = $this->getHeaderVariant();
 202                 }
 203
 204                 if ( $wgDefaultLanguageVariant && !$req ) {
 205                         $req = $this->validateVariant( $wgDefaultLanguageVariant );
 206                 }
 207
 208                 if ( $req ) {
 209                         return $req;
 210                 }
 211                 return $this->mMainLanguageCode;
 212         }
 213
 214         /**
 215          * Validate the variant and return an appropriate strict internal
 216          * variant code if one exists.  Compare to Language::hasVariant()
 217          * which does a strict test.
 218          *
 219          * @param string|null $variant The variant to validate
 220          * @return mixed Returns an equivalent valid variant code if possible,
 221          *   null otherwise
 222          */
 223         public function validateVariant( $variant = null ) {
 224                 if ( $variant === null ) {
 225                         return null;
 226                 }
 227                 // Our internal variants are always lower-case; the variant we
 228                 // are validating may have mixed case.
 229                 $variant = LanguageCode::replaceDeprecatedCodes( strtolower( $variant ) );
 230                 if ( in_array( $variant, $this->mVariants ) ) {
 231                         return $variant;
 232                 }
 233                 // Browsers are supposed to use BCP 47 standard in the
 234                 // Accept-Language header, but not all of our internal
 235                 // mediawiki variant codes are BCP 47.  Map BCP 47 code
 236                 // to our internal code.
 237                 foreach ( $this->mVariants as $v ) {
 238                         // Case-insensitive match (BCP 47 is mixed case)
 239                         if ( strtolower( LanguageCode::bcp47( $v ) ) === $variant ) {
 240                                 return $v;
 241                         }
 242                 }
 243                 return null;
 244         }
 245
 246         /**
 247          * Get the variant specified in the URL
 248          *
 249          * @return mixed Variant if one found, null otherwise
 250          */
 251         public function getURLVariant() {
 252                 global $wgRequest;
 253
 254                 if ( $this->mURLVariant ) {
 255                         return $this->mURLVariant;
 256                 }
 257
 258                 // see if the preference is set in the request
 259                 $ret = $wgRequest->getText( 'variant' );
 260
 261                 if ( !$ret ) {
 262                         $ret = $wgRequest->getVal( 'uselang' );
 263                 }
 264
 265                 $this->mURLVariant = $this->validateVariant( $ret );
 266                 return $this->mURLVariant;
 267         }
 268
 269         /**
 270          * Determine if the user has a variant set.
 271          *
 272          * @return mixed Variant if one found, null otherwise
 273          */
 274         protected function getUserVariant() {
 275                 global $wgUser;
 276
 277                 // memoizing this function wreaks havoc on parserTest.php
 278                 /*
 279                 if ( $this->mUserVariant ) {
 280                         return $this->mUserVariant;
 281                 }
 282                 */
 283
 284                 // Get language variant preference from logged in users
 285                 // Don't call this on stub objects because that causes infinite
 286                 // recursion during initialisation
 287                 if ( !$wgUser->isSafeToLoad() ) {
 288                         return false;
 289                 }
 290                 if ( $wgUser->isLoggedIn() ) {
 291                         if (
 292                                 $this->mMainLanguageCode ==
 293                                 MediaWikiServices::getInstance()->getContentLanguage()->getCode()
 294                         ) {
 295                                 $ret = $wgUser->getOption( 'variant' );
 296                         } else {
 297                                 $ret = $wgUser->getOption( 'variant-' . $this->mMainLanguageCode );
 298                         }
 299                 } else {
 300                         // figure out user lang without constructing wgLang to avoid
 301                         // infinite recursion
 302                         $ret = $wgUser->getOption( 'language' );
 303                 }
 304
 305                 $this->mUserVariant = $this->validateVariant( $ret );
 306                 return $this->mUserVariant;
 307         }
 308
 309         /**
 310          * Determine the language variant from the Accept-Language header.
 311          *
 312          * @return mixed Variant if one found, null otherwise
 313          */
 314         protected function getHeaderVariant() {
 315                 global $wgRequest;
 316
 317                 if ( $this->mHeaderVariant ) {
 318                         return $this->mHeaderVariant;
 319                 }
 320
 321                 // See if some supported language variant is set in the
 322                 // HTTP header.
 323                 $languages = array_keys( $wgRequest->getAcceptLang() );
 324                 if ( empty( $languages ) ) {
 325                         return null;
 326                 }
 327
 328                 $fallbackLanguages = [];
 329                 foreach ( $languages as $language ) {
 330                         $this->mHeaderVariant = $this->validateVariant( $language );
 331                         if ( $this->mHeaderVariant ) {
 332                                 break;
 333                         }
 334
 335                         // To see if there are fallbacks of current language.
 336                         // We record these fallback variants, and process
 337                         // them later.
 338                         $fallbacks = $this->getVariantFallbacks( $language );
 339                         if ( is_string( $fallbacks ) && $fallbacks !== $this->mMainLanguageCode ) {
 340                                 $fallbackLanguages[] = $fallbacks;
 341                         } elseif ( is_array( $fallbacks ) ) {
 342                                 $fallbackLanguages =
 343                                         array_merge( $fallbackLanguages, $fallbacks );
 344                         }
 345                 }
 346
 347                 if ( !$this->mHeaderVariant ) {
 348                         // process fallback languages now
 349                         $fallback_languages = array_unique( $fallbackLanguages );
 350                         foreach ( $fallback_languages as $language ) {
 351                                 $this->mHeaderVariant = $this->validateVariant( $language );
 352                                 if ( $this->mHeaderVariant ) {
 353                                         break;
 354                                 }
 355                         }
 356                 }
 357
 358                 return $this->mHeaderVariant;
 359         }
 360
 361         /**
 362          * Dictionary-based conversion.
 363          * This function would not parse the conversion rules.
 364          * If you want to parse rules, try to use convert() or
 365          * convertTo().
 366          *
 367          * @param string $text The text to be converted
 368          * @param bool|string $toVariant The target language code
 369          * @return string The converted text
 370          */
 371         public function autoConvert( $text, $toVariant = false ) {
 372                 $this->loadTables();
 373
 374                 if ( !$toVariant ) {
 375                         $toVariant = $this->getPreferredVariant();
 376                         if ( !$toVariant ) {
 377                                 return $text;
 378                         }
 379                 }
 380
 381                 if ( $this->guessVariant( $text, $toVariant ) ) {
 382                         return $text;
 383                 }
 384                 /* we convert everything except:
 385                    1. HTML markups (anything between < and >)
 386                    2. HTML entities
 387                    3. placeholders created by the parser
 388                    IMPORTANT: Beware of failure from pcre.backtrack_limit (T124404).
 389                    Minimize use of backtracking where possible.
 390                 */
 391                 $marker = '|' . Parser::MARKER_PREFIX . '[^\x7f]++\x7f';
 392
 393                 // this one is needed when the text is inside an HTML markup
 394                 $htmlfix = '|<[^>\004]++(?=\004$)|^[^<>]*+>';
 395
 396                 // Optimize for the common case where these tags have
 397                 // few or no children. Thus try and possesively get as much as
 398                 // possible, and only engage in backtracking when we hit a '<'.
 399
 400                 // disable convert to variants between <code> tags
 401                 $codefix = '<code>[^<]*+(?:(?:(?!<\/code>).)[^<]*+)*+<\/code>|';
 402                 // disable conversion of <script> tags
 403                 $scriptfix = '<script[^>]*+>[^<]*+(?:(?:(?!<\/script>).)[^<]*+)*+<\/script>|';
 404                 // disable conversion of <pre> tags
 405                 $prefix = '<pre[^>]*+>[^<]*+(?:(?:(?!<\/pre>).)[^<]*+)*+<\/pre>|';
 406                 // The "|.*+)" at the end, is in case we missed some part of html syntax,
 407                 // we will fail securely (hopefully) by matching the rest of the string.
 408                 $htmlFullTag = '<(?:[^>=]*+(?>[^>=]*+=\s*+(?:"[^"]*"|\'[^\']*\'|[^\'">\s]*+))*+[^>=]*+>|.*+)|';
 409
 410                 $reg = '/' . $codefix . $scriptfix . $prefix . $htmlFullTag .
 411                         '&[a-zA-Z#][a-z0-9]++;' . $marker . $htmlfix . '|\004$/s';
 412                 $startPos = 0;
 413                 $sourceBlob = '';
 414                 $literalBlob = '';
 415
 416                 // Guard against delimiter nulls in the input
 417                 // (should never happen: see T159174)
 418                 $text = str_replace( "\000", '', $text );
 419                 $text = str_replace( "\004", '', $text );
 420
 421                 $markupMatches = null;
 422                 $elementMatches = null;
 423
 424                 // We add a marker (\004) at the end of text, to ensure we always match the
 425                 // entire text (Otherwise, pcre.backtrack_limit might cause silent failure)
 426                 while ( $startPos < strlen( $text ) ) {
 427                         if ( preg_match( $reg, $text . "\004", $markupMatches, PREG_OFFSET_CAPTURE, $startPos ) ) {
 428                                 $elementPos = $markupMatches[0][1];
 429                                 $element = $markupMatches[0][0];
 430                                 if ( $element === "\004" ) {
 431                                         // We hit the end.
 432                                         $elementPos = strlen( $text );
 433                                         $element = '';
 434                                 } elseif ( substr( $element, -1 ) === "\004" ) {
 435                                         // This can sometimes happen if we have
 436                                         // unclosed html tags (For example
 437                                         // when converting a title attribute
 438                                         // during a recursive call that contains
 439                                         // a &lt; e.g. <div title="&lt;">.
 440                                         $element = substr( $element, 0, -1 );
 441                                 }
 442                         } else {
 443                                 // If we hit here, then Language Converter could be tricked
 444                                 // into doing an XSS, so we refuse to translate.
 445                                 // If non-crazy input manages to reach this code path,
 446                                 // we should consider it a bug.
 447                                 $log = LoggerFactory::getInstance( 'languageconverter' );
 448                                 $log->error( "Hit pcre.backtrack_limit in " . __METHOD__
 449                                         . ". Disabling language conversion for this page.",
 450                                         [
 451                                                 "method" => __METHOD__,
 452                                                 "variant" => $toVariant,
 453                                                 "startOfText" => substr( $text, 0, 500 )
 454                                         ]
 455                                 );
 456                                 return $text;
 457                         }
 458                         // Queue the part before the markup for translation in a batch
 459                         $sourceBlob .= substr( $text, $startPos, $elementPos - $startPos ) . "\000";
 460
 461                         // Advance to the next position
 462                         $startPos = $elementPos + strlen( $element );
 463
 464                         // Translate any alt or title attributes inside the matched element
 465                         if ( $element !== ''
 466                                 && preg_match( '/^(<[^>\s]*+)\s([^>]*+)(.*+)$/', $element, $elementMatches )
 467                         ) {
 468                                 // FIXME, this decodes entities, so if you have something
 469                                 // like <div title="foo&lt;bar"> the bar won't get
 470                                 // translated since after entity decoding it looks like
 471                                 // unclosed html and we call this method recursively
 472                                 // on attributes.
 473                                 $attrs = Sanitizer::decodeTagAttributes( $elementMatches[2] );
 474                                 // Ensure self-closing tags stay self-closing.
 475                                 $close = substr( $elementMatches[2], -1 ) === '/' ? ' /' : '';
 476                                 $changed = false;
 477                                 foreach ( [ 'title', 'alt' ] as $attrName ) {
 478                                         if ( !isset( $attrs[$attrName] ) ) {
 479                                                 continue;
 480                                         }
 481                                         $attr = $attrs[$attrName];
 482                                         // Don't convert URLs
 483                                         if ( !strpos( $attr, '://' ) ) {
 484                                                 $attr = $this->recursiveConvertTopLevel( $attr, $toVariant );
 485                                         }
 486
 487                                         if ( $attr !== $attrs[$attrName] ) {
 488                                                 $attrs[$attrName] = $attr;
 489                                                 $changed = true;
 490                                         }
 491                                 }
 492                                 if ( $changed ) {
 493                                         $element = $elementMatches[1] . Html::expandAttributes( $attrs ) .
 494                                                 $close . $elementMatches[3];
 495                                 }
 496                         }
 497                         $literalBlob .= $element . "\000";
 498                 }
 499
 500                 // Do the main translation batch
 501                 $translatedBlob = $this->translate( $sourceBlob, $toVariant );
 502
 503                 // Put the output back together
 504                 $translatedIter = StringUtils::explode( "\000", $translatedBlob );
 505                 $literalIter = StringUtils::explode( "\000", $literalBlob );
 506                 $output = '';
 507                 while ( $translatedIter->valid() && $literalIter->valid() ) {
 508                         $output .= $translatedIter->current();
 509                         $output .= $literalIter->current();
 510                         $translatedIter->next();
 511                         $literalIter->next();
 512                 }
 513
 514                 return $output;
 515         }
 516
 517         /**
 518          * Translate a string to a variant.
 519          * Doesn't parse rules or do any of that other stuff, for that use
 520          * convert() or convertTo().
 521          *
 522          * @param string $text Text to convert
 523          * @param string $variant Variant language code
 524          * @return string Translated text
 525          */
 526         public function translate( $text, $variant ) {
 527                 // If $text is empty or only includes spaces, do nothing
 528                 // Otherwise translate it
 529                 if ( trim( $text ) ) {
 530                         $this->loadTables();
 531                         $text = $this->mTables[$variant]->replace( $text );
 532                 }
 533                 return $text;
 534         }
 535
 536         /**
 537          * Call translate() to convert text to all valid variants.
 538          *
 539          * @param string $text The text to be converted
 540          * @return array Variant => converted text
 541          */
 542         public function autoConvertToAllVariants( $text ) {
 543                 $this->loadTables();
 544
 545                 $ret = [];
 546                 foreach ( $this->mVariants as $variant ) {
 547                         $ret[$variant] = $this->translate( $text, $variant );
 548                 }
 549
 550                 return $ret;
 551         }
 552
 553         /**
 554          * Apply manual conversion rules.
 555          *
 556          * @param ConverterRule $convRule
 557          */
 558         protected function applyManualConv( $convRule ) {
 559                 // Use syntax -{T|zh-cn:TitleCN; zh-tw:TitleTw}- to custom
 560                 // title conversion.
 561                 // T26072: $mConvRuleTitle was overwritten by other manual
 562                 // rule(s) not for title, this breaks the title conversion.
 563                 $newConvRuleTitle = $convRule->getTitle();
 564                 if ( $newConvRuleTitle ) {
 565                         // So I add an empty check for getTitle()
 566                         $this->mConvRuleTitle = $newConvRuleTitle;
 567                 }
 568
 569                 // merge/remove manual conversion rules to/from global table
 570                 $convTable = $convRule->getConvTable();
 571                 $action = $convRule->getRulesAction();
 572                 foreach ( $convTable as $variant => $pair ) {
 573                         $v = $this->validateVariant( $variant );
 574                         if ( !$v ) {
 575                                 continue;
 576                         }
 577
 578                         if ( $action == 'add' ) {
 579                                 // More efficient than array_merge(), about 2.5 times.
 580                                 foreach ( $pair as $from => $to ) {
 581                                         $this->mTables[$v]->setPair( $from, $to );
 582                                 }
 583                         } elseif ( $action == 'remove' ) {
 584                                 $this->mTables[$v]->removeArray( $pair );
 585                         }
 586                 }
 587         }
 588
 589         /**
 590          * Auto convert a Title object to a readable string in the
 591          * preferred variant.
 592          *
 593          * @param Title $title A object of Title
 594          * @return string Converted title text
 595          */
 596         public function convertTitle( $title ) {
 597                 $variant = $this->getPreferredVariant();
 598                 $index = $title->getNamespace();
 599                 if ( $index !== NS_MAIN ) {
 600                         $text = $this->convertNamespace( $index, $variant ) . ':';
 601                 } else {
 602                         $text = '';
 603                 }
 604                 $text .= $this->translate( $title->getText(), $variant );
 605                 return $text;
 606         }
 607
 608         /**
 609          * Get the namespace display name in the preferred variant.
 610          *
 611          * @param int $index Namespace id
 612          * @param string|null $variant Variant code or null for preferred variant
 613          * @return string Namespace name for display
 614          */
 615         public function convertNamespace( $index, $variant = null ) {
 616                 if ( $index === NS_MAIN ) {
 617                         return '';
 618                 }
 619
 620                 if ( $variant === null ) {
 621                         $variant = $this->getPreferredVariant();
 622                 }
 623
 624                 $cache = MediaWikiServices::getInstance()->getLocalServerObjectCache();
 625                 $key = $cache->makeKey( 'languageconverter', 'namespace-text', $index, $variant );
 626                 $nsVariantText = $cache->get( $key );
 627                 if ( $nsVariantText !== false ) {
 628                         return $nsVariantText;
 629                 }
 630
 631                 // First check if a message gives a converted name in the target variant.
 632                 $nsConvMsg = wfMessage( 'conversion-ns' . $index )->inLanguage( $variant );
 633                 if ( $nsConvMsg->exists() ) {
 634                         $nsVariantText = $nsConvMsg->plain();
 635                 }
 636
 637                 // Then check if a message gives a converted name in content language
 638                 // which needs extra translation to the target variant.
 639                 if ( $nsVariantText === false ) {
 640                         $nsConvMsg = wfMessage( 'conversion-ns' . $index )->inContentLanguage();
 641                         if ( $nsConvMsg->exists() ) {
 642                                 $nsVariantText = $this->translate( $nsConvMsg->plain(), $variant );
 643                         }
 644                 }
 645
 646                 if ( $nsVariantText === false ) {
 647                         // No message exists, retrieve it from the target variant's namespace names.
 648                         $langObj = $this->mLangObj->factory( $variant );
 649                         $nsVariantText = $langObj->getFormattedNsText( $index );
 650                 }
 651
 652                 $cache->set( $key, $nsVariantText, 60 );
 653
 654                 return $nsVariantText;
 655         }
 656
 657         /**
 658          * Convert text to different variants of a language. The automatic
 659          * conversion is done in autoConvert(). Here we parse the text
 660          * marked with -{}-, which specifies special conversions of the
 661          * text that can not be accomplished in autoConvert().
 662          *
 663          * Syntax of the markup:
 664          * -{code1:text1;code2:text2;...}-  or
 665          * -{flags|code1:text1;code2:text2;...}-  or
 666          * -{text}- in which case no conversion should take place for text
 667          *
 668          * @warning Glossary state is maintained between calls. Never feed this
 669          *   method input that hasn't properly been escaped as it may result in
 670          *   an XSS in subsequent calls, even if those subsequent calls properly
 671          *   escape things.
 672          * @param string $text Text to be converted, already html escaped.
 673          * @return string Converted text (html)
 674          */
 675         public function convert( $text ) {
 676                 $variant = $this->getPreferredVariant();
 677                 return $this->convertTo( $text, $variant );
 678         }
 679
 680         /**
 681          * Same as convert() except a extra parameter to custom variant.
 682          *
 683          * @param string $text Text to be converted, already html escaped
 684          * @param-taint $text exec_html
 685          * @param string $variant The target variant code
 686          * @return string Converted text
 687          * @return-taint escaped
 688          */
 689         public function convertTo( $text, $variant ) {
 690                 global $wgDisableLangConversion;
 691                 if ( $wgDisableLangConversion ) {
 692                         return $text;
 693                 }
 694                 // Reset converter state for a new converter run.
 695                 $this->mConvRuleTitle = false;
 696                 return $this->recursiveConvertTopLevel( $text, $variant );
 697         }
 698
 699         /**
 700          * Recursively convert text on the outside. Allow to use nested
 701          * markups to custom rules.
 702          *
 703          * @param string $text Text to be converted
 704          * @param string $variant The target variant code
 705          * @param int $depth Depth of recursion
 706          * @return string Converted text
 707          */
 708         protected function recursiveConvertTopLevel( $text, $variant, $depth = 0 ) {
 709                 $startPos = 0;
 710                 $out = '';
 711                 $length = strlen( $text );
 712                 $shouldConvert = !$this->guessVariant( $text, $variant );
 713                 $continue = 1;
 714
 715                 $noScript = '<script.*?>.*?<\/script>(*SKIP)(*FAIL)';
 716                 $noStyle = '<style.*?>.*?<\/style>(*SKIP)(*FAIL)';
 717                 // phpcs:ignore Generic.Files.LineLength
 718                 $noHtml = '<(?:[^>=]*+(?>[^>=]*+=\s*+(?:"[^"]*"|\'[^\']*\'|[^\'">\s]*+))*+[^>=]*+>|.*+)(*SKIP)(*FAIL)';
 719                 while ( $startPos < $length && $continue ) {
 720                         $continue = preg_match(
 721                                 // Only match -{ outside of html.
 722                                 "/$noScript|$noStyle|$noHtml|-\{/",
 723                                 $text,
 724                                 $m,
 725                                 PREG_OFFSET_CAPTURE,
 726                                 $startPos
 727                         );
 728
 729                         if ( !$continue ) {
 730                                 // No more markup, append final segment
 731                                 $fragment = substr( $text, $startPos );
 732                                 $out .= $shouldConvert ? $this->autoConvert( $fragment, $variant ) : $fragment;
 733                                 return $out;
 734                         }
 735
 736                         // Offset of the match of the regex pattern.
 737                         $pos = $m[0][1];
 738
 739                         // Append initial segment
 740                         $fragment = substr( $text, $startPos, $pos - $startPos );
 741                         $out .= $shouldConvert ? $this->autoConvert( $fragment, $variant ) : $fragment;
 742                         // -{ marker found, not in attribute
 743                         // Advance position up to -{ marker.
 744                         $startPos = $pos;
 745                         // Do recursive conversion
 746                         // Note: This passes $startPos by reference, and advances it.
 747                         $out .= $this->recursiveConvertRule( $text, $variant, $startPos, $depth + 1 );
 748                 }
 749                 return $out;
 750         }
 751
 752         /**
 753          * Recursively convert text on the inside.
 754          *
 755          * @param string $text Text to be converted
 756          * @param string $variant The target variant code
 757          * @param int &$startPos
 758          * @param int $depth Depth of recursion
 759          *
 760          * @throws MWException
 761          * @return string Converted text
 762          */
 763         protected function recursiveConvertRule( $text, $variant, &$startPos, $depth = 0 ) {
 764                 // Quick sanity check (no function calls)
 765                 if ( $text[$startPos] !== '-' || $text[$startPos + 1] !== '{' ) {
 766                         throw new MWException( __METHOD__ . ': invalid input string' );
 767                 }
 768
 769                 $startPos += 2;
 770                 $inner = '';
 771                 $warningDone = false;
 772                 $length = strlen( $text );
 773
 774                 while ( $startPos < $length ) {
 775                         $m = false;
 776                         preg_match( '/-\{|\}-/', $text, $m, PREG_OFFSET_CAPTURE, $startPos );
 777                         if ( !$m ) {
 778                                 // Unclosed rule
 779                                 break;
 780                         }
 781
 782                         $token = $m[0][0];
 783                         $pos = $m[0][1];
 784
 785                         // Markup found
 786                         // Append initial segment
 787                         $inner .= substr( $text, $startPos, $pos - $startPos );
 788
 789                         // Advance position
 790                         $startPos = $pos;
 791
 792                         switch ( $token ) {
 793                                 case '-{':
 794                                         // Check max depth
 795                                         if ( $depth >= $this->mMaxDepth ) {
 796                                                 $inner .= '-{';
 797                                                 if ( !$warningDone ) {
 798                                                         $inner .= '<span class="error">' .
 799                                                                 wfMessage( 'language-converter-depth-warning' )
 800                                                                         ->numParams( $this->mMaxDepth )->inContentLanguage()->text() .
 801                                                                 '</span>';
 802                                                         $warningDone = true;
 803                                                 }
 804                                                 $startPos += 2;
 805                                                 break;
 806                                         }
 807                                         // Recursively parse another rule
 808                                         $inner .= $this->recursiveConvertRule( $text, $variant, $startPos, $depth + 1 );
 809                                         break;
 810                                 case '}-':
 811                                         // Apply the rule
 812                                         $startPos += 2;
 813                                         $rule = new ConverterRule( $inner, $this );
 814                                         $rule->parse( $variant );
 815                                         $this->applyManualConv( $rule );
 816                                         return $rule->getDisplay();
 817                                 default:
 818                                         throw new MWException( __METHOD__ . ': invalid regex match' );
 819                         }
 820                 }
 821
 822                 // Unclosed rule
 823                 if ( $startPos < $length ) {
 824                         $inner .= substr( $text, $startPos );
 825                 }
 826                 $startPos = $length;
 827                 return '-{' . $this->autoConvert( $inner, $variant );
 828         }
 829
 830         /**
 831          * If a language supports multiple variants, it is possible that
 832          * non-existing link in one variant actually exists in another variant.
 833          * This function tries to find it. See e.g. LanguageZh.php
 834          * The input parameters may be modified upon return
 835          *
 836          * @param string &$link The name of the link
 837          * @param Title &$nt The title object of the link
 838          * @param bool $ignoreOtherCond To disable other conditions when
 839          *   we need to transclude a template or update a category's link
 840          */
 841         public function findVariantLink( &$link, &$nt, $ignoreOtherCond = false ) {
 842                 # If the article has already existed, there is no need to
 843                 # check it again, otherwise it may cause a fault.
 844                 if ( is_object( $nt ) && $nt->exists() ) {
 845                         return;
 846                 }
 847
 848                 global $wgDisableLangConversion, $wgDisableTitleConversion, $wgRequest;
 849                 $isredir = $wgRequest->getText( 'redirect', 'yes' );
 850                 $action = $wgRequest->getText( 'action' );
 851                 if ( $action == 'edit' && $wgRequest->getBool( 'redlink' ) ) {
 852                         $action = 'view';
 853                 }
 854                 $linkconvert = $wgRequest->getText( 'linkconvert', 'yes' );
 855                 $disableLinkConversion = $wgDisableLangConversion
 856                         || $wgDisableTitleConversion;
 857                 $linkBatch = new LinkBatch();
 858
 859                 $ns = NS_MAIN;
 860
 861                 if ( $disableLinkConversion ||
 862                         ( !$ignoreOtherCond &&
 863                                 ( $isredir == 'no'
 864                                         || $action == 'edit'
 865                                         || $action == 'submit'
 866                                         || $linkconvert == 'no' ) ) ) {
 867                         return;
 868                 }
 869
 870                 if ( is_object( $nt ) ) {
 871                         $ns = $nt->getNamespace();
 872                 }
 873
 874                 $variants = $this->autoConvertToAllVariants( $link );
 875                 if ( !$variants ) { // give up
 876                         return;
 877                 }
 878
 879                 $titles = [];
 880
 881                 foreach ( $variants as $v ) {
 882                         if ( $v != $link ) {
 883                                 $varnt = Title::newFromText( $v, $ns );
 884                                 if ( !is_null( $varnt ) ) {
 885                                         $linkBatch->addObj( $varnt );
 886                                         $titles[] = $varnt;
 887                                 }
 888                         }
 889                 }
 890
 891                 // fetch all variants in single query
 892                 $linkBatch->execute();
 893
 894                 foreach ( $titles as $varnt ) {
 895                         if ( $varnt->getArticleID() > 0 ) {
 896                                 $nt = $varnt;
 897                                 $link = $varnt->getText();
 898                                 break;
 899                         }
 900                 }
 901         }
 902
 903         /**
 904          * Returns language specific hash options.
 905          *
 906          * @return string
 907          */
 908         public function getExtraHashOptions() {
 909                 $variant = $this->getPreferredVariant();
 910
 911                 return '!' . $variant;
 912         }
 913
 914         /**
 915          * Guess if a text is written in a variant. This should be implemented in subclasses.
 916          *
 917          * @param string $text The text to be checked
 918          * @param string $variant Language code of the variant to be checked for
 919          * @return bool True if $text appears to be written in $variant, false if not
 920          *
 921          * @author Nikola Smolenski <smolensk@eunet.rs>
 922          * @since 1.19
 923          */
 924         public function guessVariant( $text, $variant ) {
 925                 return false;
 926         }
 927
 928         /**
 929          * Load default conversion tables.
 930          * This method must be implemented in derived class.
 931          *
 932          * @private
 933          * @throws MWException
 934          */
 935         function loadDefaultTables() {
 936                 $class = static::class;
 937                 throw new MWException( "Must implement loadDefaultTables() method in class $class" );
 938         }
 939
 940         /**
 941          * Load conversion tables either from the cache or the disk.
 942          * @private
 943          * @param bool $fromCache Load from memcached? Defaults to true.
 944          */
 945         function loadTables( $fromCache = true ) {
 946                 global $wgLanguageConverterCacheType;
 947
 948                 if ( $this->mTablesLoaded ) {
 949                         return;
 950                 }
 951
 952                 $this->mTablesLoaded = true;
 953                 $this->mTables = false;
 954                 $cache = ObjectCache::getInstance( $wgLanguageConverterCacheType );
 955                 $cacheKey = $cache->makeKey( 'conversiontables', $this->mMainLanguageCode );
 956                 if ( $fromCache ) {
 957                         $this->mTables = $cache->get( $cacheKey );
 958                 }
 959                 if ( !$this->mTables || !array_key_exists( self::CACHE_VERSION_KEY, $this->mTables ) ) {
 960                         // not in cache, or we need a fresh reload.
 961                         // We will first load the default tables
 962                         // then update them using things in MediaWiki:Conversiontable/*
 963                         $this->loadDefaultTables();
 964                         foreach ( $this->mVariants as $var ) {
 965                                 $cached = $this->parseCachedTable( $var );
 966                                 $this->mTables[$var]->mergeArray( $cached );
 967                         }
 968
 969                         $this->postLoadTables();
 970                         $this->mTables[self::CACHE_VERSION_KEY] = true;
 971
 972                         $cache->set( $cacheKey, $this->mTables, 43200 );
 973                 }
 974         }
 975
 976         /**
 977          * Hook for post processing after conversion tables are loaded.
 978          */
 979         function postLoadTables() {
 980         }
 981
 982         /**
 983          * Reload the conversion tables.
 984          *
 985          * Also used by test suites which need to reset the converter state.
 986          *
 987          * @private
 988          */
 989         private function reloadTables() {
 990                 if ( $this->mTables ) {
 991                         unset( $this->mTables );
 992                 }
 993
 994                 $this->mTablesLoaded = false;
 995                 $this->loadTables( false );
 996         }
 997
 998         /**
 999          * Parse the conversion table stored in the cache.
1000          *
1001          * The tables should be in blocks of the following form:
1002          *              -{
1003          *                      word => word ;
1004          *                      word => word ;
1005          *                      ...
1006          *              }-
1007          *
1008          * To make the tables more manageable, subpages are allowed
1009          * and will be parsed recursively if $recursive == true.
1010          *
1011          * @param string $code Language code
1012          * @param string $subpage Subpage name
1013          * @param bool $recursive Parse subpages recursively? Defaults to true.
1014          *
1015          * @return array
1016          */
1017         function parseCachedTable( $code, $subpage = '', $recursive = true ) {
1018                 static $parsed = [];
1019
1020                 $key = 'Conversiontable/' . $code;
1021                 if ( $subpage ) {
1022                         $key .= '/' . $subpage;
1023                 }
1024                 if ( array_key_exists( $key, $parsed ) ) {
1025                         return [];
1026                 }
1027
1028                 $parsed[$key] = true;
1029
1030                 if ( $subpage === '' ) {
1031                         $txt = MessageCache::singleton()->getMsgFromNamespace( $key, $code );
1032                 } else {
1033                         $txt = false;
1034                         $title = Title::makeTitleSafe( NS_MEDIAWIKI, $key );
1035                         if ( $title && $title->exists() ) {
1036                                 $revision = Revision::newFromTitle( $title );
1037                                 if ( $revision ) {
1038                                         if ( $revision->getContentModel() == CONTENT_MODEL_WIKITEXT ) {
1039                                                 $txt = $revision->getContent( Revision::RAW )->getNativeData();
1040                                         }
1041
1042                                         // @todo in the future, use a specialized content model, perhaps based on json!
1043                                 }
1044                         }
1045                 }
1046
1047                 # Nothing to parse if there's no text
1048                 if ( $txt === false || $txt === null || $txt === '' ) {
1049                         return [];
1050                 }
1051
1052                 // get all subpage links of the form
1053                 // [[MediaWiki:Conversiontable/zh-xx/...|...]]
1054                 $linkhead = $this->mLangObj->getNsText( NS_MEDIAWIKI ) .
1055                         ':Conversiontable';
1056                 $subs = StringUtils::explode( '[[', $txt );
1057                 $sublinks = [];
1058                 foreach ( $subs as $sub ) {
1059                         $link = explode( ']]', $sub, 2 );
1060                         if ( count( $link ) != 2 ) {
1061                                 continue;
1062                         }
1063                         $b = explode( '|', $link[0], 2 );
1064                         $b = explode( '/', trim( $b[0] ), 3 );
1065                         if ( count( $b ) == 3 ) {
1066                                 $sublink = $b[2];
1067                         } else {
1068                                 $sublink = '';
1069                         }
1070
1071                         if ( $b[0] == $linkhead && $b[1] == $code ) {
1072                                 $sublinks[] = $sublink;
1073                         }
1074                 }
1075
1076                 // parse the mappings in this page
1077                 $blocks = StringUtils::explode( '-{', $txt );
1078                 $ret = [];
1079                 $first = true;
1080                 foreach ( $blocks as $block ) {
1081                         if ( $first ) {
1082                                 // Skip the part before the first -{
1083                                 $first = false;
1084                                 continue;
1085                         }
1086                         $mappings = explode( '}-', $block, 2 )[0];
1087                         $stripped = str_replace( [ "'", '"', '*', '#' ], '', $mappings );
1088                         $table = StringUtils::explode( ';', $stripped );
1089                         foreach ( $table as $t ) {
1090                                 $m = explode( '=>', $t, 3 );
1091                                 if ( count( $m ) != 2 ) {
1092                                         continue;
1093                                 }
1094                                 // trim any trailling comments starting with '//'
1095                                 $tt = explode( '//', $m[1], 2 );
1096                                 $ret[trim( $m[0] )] = trim( $tt[0] );
1097                         }
1098                 }
1099
1100                 // recursively parse the subpages
1101                 if ( $recursive ) {
1102                         foreach ( $sublinks as $link ) {
1103                                 $s = $this->parseCachedTable( $code, $link, $recursive );
1104                                 $ret = $s + $ret;
1105                         }
1106                 }
1107
1108                 if ( $this->mUcfirst ) {
1109                         foreach ( $ret as $k => $v ) {
1110                                 $ret[$this->mLangObj->ucfirst( $k )] = $this->mLangObj->ucfirst( $v );
1111                         }
1112                 }
1113                 return $ret;
1114         }
1115
1116         /**
1117          * Enclose a string with the "no conversion" tag. This is used by
1118          * various functions in the Parser.
1119          *
1120          * @param string $text Text to be tagged for no conversion
1121          * @param bool $noParse Unused
1122          * @return string The tagged text
1123          */
1124         public function markNoConversion( $text, $noParse = false ) {
1125                 # don't mark if already marked
1126                 if ( strpos( $text, '-{' ) || strpos( $text, '}-' ) ) {
1127                         return $text;
1128                 }
1129
1130                 $ret = "-{R|$text}-";
1131                 return $ret;
1132         }
1133
1134         /**
1135          * Convert the sorting key for category links. This should make different
1136          * keys that are variants of each other map to the same key.
1137          *
1138          * @param string $key
1139          *
1140          * @return string
1141          */
1142         function convertCategoryKey( $key ) {
1143                 return $key;
1144         }
1145
1146         /**
1147          * Refresh the cache of conversion tables when
1148          * MediaWiki:Conversiontable* is updated.
1149          *
1150          * @param Title $titleobj The Title of the page being updated
1151          */
1152         public function updateConversionTable( Title $titleobj ) {
1153                 if ( $titleobj->getNamespace() == NS_MEDIAWIKI ) {
1154                         $title = $titleobj->getDBkey();
1155                         $t = explode( '/', $title, 3 );
1156                         $c = count( $t );
1157                         if ( $c > 1 && $t[0] == 'Conversiontable' ) {
1158                                 if ( $this->validateVariant( $t[1] ) ) {
1159                                         $this->reloadTables();
1160                                 }
1161                         }
1162                 }
1163         }
1164
1165         /**
1166          * Get the cached separator pattern for ConverterRule::parseRules()
1167          * @return string
1168          */
1169         function getVarSeparatorPattern() {
1170                 if ( is_null( $this->mVarSeparatorPattern ) ) {
1171                         // varsep_pattern for preg_split:
1172                         // text should be splited by ";" only if a valid variant
1173                         // name exist after the markup, for example:
1174                         //  -{zh-hans:<span style="font-size:120%;">xxx</span>;zh-hant:\
1175                         //  <span style="font-size:120%;">yyy</span>;}-
1176                         // we should split it as:
1177                         //  [
1178                         //    [0] => 'zh-hans:<span style="font-size:120%;">xxx</span>'
1179                         //    [1] => 'zh-hant:<span style="font-size:120%;">yyy</span>'
1180                         //    [2] => ''
1181                         //  ]
1182                         $pat = '/;\s*(?=';
1183                         foreach ( $this->mVariants as $variant ) {
1184                                 // zh-hans:xxx;zh-hant:yyy
1185                                 $pat .= $variant . '\s*:|';
1186                                 // xxx=>zh-hans:yyy; xxx=>zh-hant:zzz
1187                                 $pat .= '[^;]*?=>\s*' . $variant . '\s*:|';
1188                         }
1189                         $pat .= '\s*$)/';
1190                         $this->mVarSeparatorPattern = $pat;
1191                 }
1192                 return $this->mVarSeparatorPattern;
1193         }
1194 }