languages/LanguageConverter.php

   1 <?php
   2 /**
   3  * This program is free software; you can redistribute it and/or modify
   4  * it under the terms of the GNU General Public License as published by
   5  * the Free Software Foundation; either version 2 of the License, or
   6  * (at your option) any later version.
   7  *
   8  * This program is distributed in the hope that it will be useful,
   9  * but WITHOUT ANY WARRANTY; without even the implied warranty of
  10  * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the
  11  * GNU General Public License for more details.
  12  *
  13  * You should have received a copy of the GNU General Public License along
  14  * with this program; if not, write to the Free Software Foundation, Inc.,
  15  * 51 Franklin Street, Fifth Floor, Boston, MA 02110-1301, USA.
  16  * http://www.gnu.org/copyleft/gpl.html
  17  *
  18  * @file
  19  * @ingroup Language
  20  */
  21 use MediaWiki\MediaWikiServices;
  22
  23 use MediaWiki\Logger\LoggerFactory;
  24
  25 /**
  26  * Base class for language conversion.
  27  * @ingroup Language
  28  *
  29  * @author Zhengzhu Feng <zhengzhu@gmail.com>
  30  * @author fdcn <fdcn64@gmail.com>
  31  * @author shinjiman <shinjiman@gmail.com>
  32  * @author PhiLiP <philip.npc@gmail.com>
  33  */
  34 class LanguageConverter {
  35         /**
  36          * languages supporting variants
  37          * @since 1.20
  38          * @var array
  39          */
  40         static public $languagesWithVariants = [
  41                 'en',
  42                 'gan',
  43                 'iu',
  44                 'kk',
  45                 'ku',
  46                 'shi',
  47                 'sr',
  48                 'tg',
  49                 'uz',
  50                 'zh',
  51         ];
  52
  53         public $mMainLanguageCode;
  54
  55         /**
  56          * @var string[]
  57          */
  58         public $mVariants;
  59         public $mVariantFallbacks;
  60         public $mVariantNames;
  61         public $mTablesLoaded = false;
  62         public $mTables;
  63         // 'bidirectional' 'unidirectional' 'disable' for each variant
  64         public $mManualLevel;
  65
  66         public $mLangObj;
  67         public $mFlags;
  68         public $mDescCodeSep = ':', $mDescVarSep = ';';
  69         public $mUcfirst = false;
  70         public $mConvRuleTitle = false;
  71         public $mURLVariant;
  72         public $mUserVariant;
  73         public $mHeaderVariant;
  74         public $mMaxDepth = 10;
  75         public $mVarSeparatorPattern;
  76
  77         const CACHE_VERSION_KEY = 'VERSION 7';
  78
  79         /**
  80          * @param Language $langobj
  81          * @param string $maincode The main language code of this language
  82          * @param string[] $variants The supported variants of this language
  83          * @param array $variantfallbacks The fallback language of each variant
  84          * @param array $flags Defining the custom strings that maps to the flags
  85          * @param array $manualLevel Limit for supported variants
  86          */
  87         public function __construct( $langobj, $maincode, $variants = [],
  88                                                                 $variantfallbacks = [], $flags = [],
  89                                                                 $manualLevel = [] ) {
  90                 global $wgDisabledVariants;
  91                 $this->mLangObj = $langobj;
  92                 $this->mMainLanguageCode = $maincode;
  93                 $this->mVariants = array_diff( $variants, $wgDisabledVariants );
  94                 $this->mVariantFallbacks = $variantfallbacks;
  95                 $this->mVariantNames = Language::fetchLanguageNames();
  96                 $defaultflags = [
  97                         // 'S' show converted text
  98                         // '+' add rules for alltext
  99                         // 'E' the gave flags is error
 100                         // these flags above are reserved for program
 101                         'A' => 'A',   // add rule for convert code (all text convert)
 102                         'T' => 'T',   // title convert
 103                         'R' => 'R',   // raw content
 104                         'D' => 'D',   // convert description (subclass implement)
 105                         '-' => '-',   // remove convert (not implement)
 106                         'H' => 'H',   // add rule for convert code (but no display in placed code)
 107                         'N' => 'N',   // current variant name
 108                 ];
 109                 $this->mFlags = array_merge( $defaultflags, $flags );
 110                 foreach ( $this->mVariants as $v ) {
 111                         if ( array_key_exists( $v, $manualLevel ) ) {
 112                                 $this->mManualLevel[$v] = $manualLevel[$v];
 113                         } else {
 114                                 $this->mManualLevel[$v] = 'bidirectional';
 115                         }
 116                         $this->mFlags[$v] = $v;
 117                 }
 118         }
 119
 120         /**
 121          * Get all valid variants.
 122          * Call this instead of using $this->mVariants directly.
 123          *
 124          * @return string[] Contains all valid variants
 125          */
 126         public function getVariants() {
 127                 return $this->mVariants;
 128         }
 129
 130         /**
 131          * In case some variant is not defined in the markup, we need
 132          * to have some fallback. For example, in zh, normally people
 133          * will define zh-hans and zh-hant, but less so for zh-sg or zh-hk.
 134          * when zh-sg is preferred but not defined, we will pick zh-hans
 135          * in this case. Right now this is only used by zh.
 136          *
 137          * @param string $variant The language code of the variant
 138          * @return string|array The code of the fallback language or the
 139          *   main code if there is no fallback
 140          */
 141         public function getVariantFallbacks( $variant ) {
 142                 if ( isset( $this->mVariantFallbacks[$variant] ) ) {
 143                         return $this->mVariantFallbacks[$variant];
 144                 }
 145                 return $this->mMainLanguageCode;
 146         }
 147
 148         /**
 149          * Get the title produced by the conversion rule.
 150          * @return string The converted title text
 151          */
 152         public function getConvRuleTitle() {
 153                 return $this->mConvRuleTitle;
 154         }
 155
 156         /**
 157          * Get preferred language variant.
 158          * @return string The preferred language code
 159          */
 160         public function getPreferredVariant() {
 161                 global $wgDefaultLanguageVariant, $wgUser;
 162
 163                 $req = $this->getURLVariant();
 164
 165                 if ( $wgUser->isSafeToLoad() && $wgUser->isLoggedIn() && !$req ) {
 166                         $req = $this->getUserVariant();
 167                 } elseif ( !$req ) {
 168                         $req = $this->getHeaderVariant();
 169                 }
 170
 171                 if ( $wgDefaultLanguageVariant && !$req ) {
 172                         $req = $this->validateVariant( $wgDefaultLanguageVariant );
 173                 }
 174
 175                 // This function, unlike the other get*Variant functions, is
 176                 // not memoized (i.e. there return value is not cached) since
 177                 // new information might appear during processing after this
 178                 // is first called.
 179                 if ( $this->validateVariant( $req ) ) {
 180                         return $req;
 181                 }
 182                 return $this->mMainLanguageCode;
 183         }
 184
 185         /**
 186          * Get default variant.
 187          * This function would not be affected by user's settings
 188          * @return string The default variant code
 189          */
 190         public function getDefaultVariant() {
 191                 global $wgDefaultLanguageVariant;
 192
 193                 $req = $this->getURLVariant();
 194
 195                 if ( !$req ) {
 196                         $req = $this->getHeaderVariant();
 197                 }
 198
 199                 if ( $wgDefaultLanguageVariant && !$req ) {
 200                         $req = $this->validateVariant( $wgDefaultLanguageVariant );
 201                 }
 202
 203                 if ( $req ) {
 204                         return $req;
 205                 }
 206                 return $this->mMainLanguageCode;
 207         }
 208
 209         /**
 210          * Validate the variant
 211          * @param string $variant The variant to validate
 212          * @return mixed Returns the variant if it is valid, null otherwise
 213          */
 214         public function validateVariant( $variant = null ) {
 215                 if ( $variant !== null && in_array( $variant, $this->mVariants ) ) {
 216                         return $variant;
 217                 }
 218                 return null;
 219         }
 220
 221         /**
 222          * Get the variant specified in the URL
 223          *
 224          * @return mixed Variant if one found, false otherwise.
 225          */
 226         public function getURLVariant() {
 227                 global $wgRequest;
 228
 229                 if ( $this->mURLVariant ) {
 230                         return $this->mURLVariant;
 231                 }
 232
 233                 // see if the preference is set in the request
 234                 $ret = $wgRequest->getText( 'variant' );
 235
 236                 if ( !$ret ) {
 237                         $ret = $wgRequest->getVal( 'uselang' );
 238                 }
 239
 240                 $this->mURLVariant = $this->validateVariant( $ret );
 241                 return $this->mURLVariant;
 242         }
 243
 244         /**
 245          * Determine if the user has a variant set.
 246          *
 247          * @return mixed Variant if one found, false otherwise.
 248          */
 249         protected function getUserVariant() {
 250                 global $wgUser, $wgContLang;
 251
 252                 // memoizing this function wreaks havoc on parserTest.php
 253                 /*
 254                 if ( $this->mUserVariant ) {
 255                         return $this->mUserVariant;
 256                 }
 257                 */
 258
 259                 // Get language variant preference from logged in users
 260                 // Don't call this on stub objects because that causes infinite
 261                 // recursion during initialisation
 262                 if ( !$wgUser->isSafeToLoad() ) {
 263                         return false;
 264                 }
 265                 if ( $wgUser->isLoggedIn() ) {
 266                         if ( $this->mMainLanguageCode == $wgContLang->getCode() ) {
 267                                 $ret = $wgUser->getOption( 'variant' );
 268                         } else {
 269                                 $ret = $wgUser->getOption( 'variant-' . $this->mMainLanguageCode );
 270                         }
 271                 } else {
 272                         // figure out user lang without constructing wgLang to avoid
 273                         // infinite recursion
 274                         $ret = $wgUser->getOption( 'language' );
 275                 }
 276
 277                 $this->mUserVariant = $this->validateVariant( $ret );
 278                 return $this->mUserVariant;
 279         }
 280
 281         /**
 282          * Determine the language variant from the Accept-Language header.
 283          *
 284          * @return mixed Variant if one found, false otherwise.
 285          */
 286         protected function getHeaderVariant() {
 287                 global $wgRequest;
 288
 289                 if ( $this->mHeaderVariant ) {
 290                         return $this->mHeaderVariant;
 291                 }
 292
 293                 // see if some supported language variant is set in the
 294                 // HTTP header.
 295                 $languages = array_keys( $wgRequest->getAcceptLang() );
 296                 if ( empty( $languages ) ) {
 297                         return null;
 298                 }
 299
 300                 $fallbackLanguages = [];
 301                 foreach ( $languages as $language ) {
 302                         $this->mHeaderVariant = $this->validateVariant( $language );
 303                         if ( $this->mHeaderVariant ) {
 304                                 break;
 305                         }
 306
 307                         // To see if there are fallbacks of current language.
 308                         // We record these fallback variants, and process
 309                         // them later.
 310                         $fallbacks = $this->getVariantFallbacks( $language );
 311                         if ( is_string( $fallbacks ) && $fallbacks !== $this->mMainLanguageCode ) {
 312                                 $fallbackLanguages[] = $fallbacks;
 313                         } elseif ( is_array( $fallbacks ) ) {
 314                                 $fallbackLanguages =
 315                                         array_merge( $fallbackLanguages, $fallbacks );
 316                         }
 317                 }
 318
 319                 if ( !$this->mHeaderVariant ) {
 320                         // process fallback languages now
 321                         $fallback_languages = array_unique( $fallbackLanguages );
 322                         foreach ( $fallback_languages as $language ) {
 323                                 $this->mHeaderVariant = $this->validateVariant( $language );
 324                                 if ( $this->mHeaderVariant ) {
 325                                         break;
 326                                 }
 327                         }
 328                 }
 329
 330                 return $this->mHeaderVariant;
 331         }
 332
 333         /**
 334          * Dictionary-based conversion.
 335          * This function would not parse the conversion rules.
 336          * If you want to parse rules, try to use convert() or
 337          * convertTo().
 338          *
 339          * @param string $text The text to be converted
 340          * @param bool|string $toVariant The target language code
 341          * @return string The converted text
 342          */
 343         public function autoConvert( $text, $toVariant = false ) {
 344                 $this->loadTables();
 345
 346                 if ( !$toVariant ) {
 347                         $toVariant = $this->getPreferredVariant();
 348                         if ( !$toVariant ) {
 349                                 return $text;
 350                         }
 351                 }
 352
 353                 if ( $this->guessVariant( $text, $toVariant ) ) {
 354                         return $text;
 355                 }
 356
 357                 /* we convert everything except:
 358                    1. HTML markups (anything between < and >)
 359                    2. HTML entities
 360                    3. placeholders created by the parser
 361                    IMPORTANT: Beware of failure from pcre.backtrack_limit (T124404).
 362                    Minimize use of backtracking where possible.
 363                 */
 364                 $marker = '|' . Parser::MARKER_PREFIX . '[^\x7f]++\x7f';
 365
 366                 // this one is needed when the text is inside an HTML markup
 367                 $htmlfix = '|<[^>\004]++(?=\004$)|^[^<>]*+>';
 368
 369                 // Optimize for the common case where these tags have
 370                 // few or no children. Thus try and possesively get as much as
 371                 // possible, and only engage in backtracking when we hit a '<'.
 372
 373                 // disable convert to variants between <code> tags
 374                 $codefix = '<code>[^<]*+(?:(?:(?!<\/code>).)[^<]*+)*+<\/code>|';
 375                 // disable conversion of <script> tags
 376                 $scriptfix = '<script[^>]*+>[^<]*+(?:(?:(?!<\/script>).)[^<]*+)*+<\/script>|';
 377                 // disable conversion of <pre> tags
 378                 $prefix = '<pre[^>]*+>[^<]*+(?:(?:(?!<\/pre>).)[^<]*+)*+<\/pre>|';
 379                 // The "|.*+)" at the end, is in case we missed some part of html syntax,
 380                 // we will fail securely (hopefully) by matching the rest of the string.
 381                 $htmlFullTag = '<(?:[^>=]*+(?>[^>=]*+=\s*+(?:"[^"]*"|\'[^\']*\'|[^\'">\s]*+))*+[^>=]*+>|.*+)|';
 382
 383                 $reg = '/' . $codefix . $scriptfix . $prefix . $htmlFullTag .
 384                         '&[a-zA-Z#][a-z0-9]++;' . $marker . $htmlfix . '|\004$/s';
 385                 $startPos = 0;
 386                 $sourceBlob = '';
 387                 $literalBlob = '';
 388
 389                 // Guard against delimiter nulls in the input
 390                 // (should never happen: see T159174)
 391                 $text = str_replace( "\000", '', $text );
 392
 393                 $markupMatches = null;
 394                 $elementMatches = null;
 395
 396                 // We add a marker (\004) at the end of text, to ensure we always match the
 397                 // entire text (Otherwise, pcre.backtrack_limit might cause silent failure)
 398                 while ( $startPos < strlen( $text ) ) {
 399                         if ( preg_match( $reg, $text . "\004", $markupMatches, PREG_OFFSET_CAPTURE, $startPos ) ) {
 400                                 $elementPos = $markupMatches[0][1];
 401                                 $element = $markupMatches[0][0];
 402                                 if ( $element === "\004" ) {
 403                                         // We hit the end.
 404                                         $elementPos = strlen( $text );
 405                                         $element = '';
 406                                 }
 407                         } else {
 408                                 // If we hit here, then Language Converter could be tricked
 409                                 // into doing an XSS, so we refuse to translate.
 410                                 // If non-crazy input manages to reach this code path,
 411                                 // we should consider it a bug.
 412                                 $log = LoggerFactory::getInstance( 'languageconverter' );
 413                                 $log->error( "Hit pcre.backtrack_limit in " . __METHOD__
 414                                         . ". Disabling language conversion for this page.",
 415                                         [
 416                                                 "method" => __METHOD__,
 417                                                 "variant" => $toVariant,
 418                                                 "startOfText" => substr( $text, 0, 500 )
 419                                         ]
 420                                 );
 421                                 return $text;
 422                         }
 423                         // Queue the part before the markup for translation in a batch
 424                         $sourceBlob .= substr( $text, $startPos, $elementPos - $startPos ) . "\000";
 425
 426                         // Advance to the next position
 427                         $startPos = $elementPos + strlen( $element );
 428
 429                         // Translate any alt or title attributes inside the matched element
 430                         if ( $element !== ''
 431                                 && preg_match( '/^(<[^>\s]*+)\s([^>]*+)(.*+)$/', $element, $elementMatches )
 432                         ) {
 433                                 $attrs = Sanitizer::decodeTagAttributes( $elementMatches[2] );
 434                                 $changed = false;
 435                                 foreach ( [ 'title', 'alt' ] as $attrName ) {
 436                                         if ( !isset( $attrs[$attrName] ) ) {
 437                                                 continue;
 438                                         }
 439                                         $attr = $attrs[$attrName];
 440                                         // Don't convert URLs
 441                                         if ( !strpos( $attr, '://' ) ) {
 442                                                 $attr = $this->recursiveConvertTopLevel( $attr, $toVariant );
 443                                         }
 444
 445                                         if ( $attr !== $attrs[$attrName] ) {
 446                                                 $attrs[$attrName] = $attr;
 447                                                 $changed = true;
 448                                         }
 449                                 }
 450                                 if ( $changed ) {
 451                                         $element = $elementMatches[1] . Html::expandAttributes( $attrs ) .
 452                                                 $elementMatches[3];
 453                                 }
 454                         }
 455                         $literalBlob .= $element . "\000";
 456                 }
 457
 458                 // Do the main translation batch
 459                 $translatedBlob = $this->translate( $sourceBlob, $toVariant );
 460
 461                 // Put the output back together
 462                 $translatedIter = StringUtils::explode( "\000", $translatedBlob );
 463                 $literalIter = StringUtils::explode( "\000", $literalBlob );
 464                 $output = '';
 465                 while ( $translatedIter->valid() && $literalIter->valid() ) {
 466                         $output .= $translatedIter->current();
 467                         $output .= $literalIter->current();
 468                         $translatedIter->next();
 469                         $literalIter->next();
 470                 }
 471
 472                 return $output;
 473         }
 474
 475         /**
 476          * Translate a string to a variant.
 477          * Doesn't parse rules or do any of that other stuff, for that use
 478          * convert() or convertTo().
 479          *
 480          * @param string $text Text to convert
 481          * @param string $variant Variant language code
 482          * @return string Translated text
 483          */
 484         public function translate( $text, $variant ) {
 485                 // If $text is empty or only includes spaces, do nothing
 486                 // Otherwise translate it
 487                 if ( trim( $text ) ) {
 488                         $this->loadTables();
 489                         $text = $this->mTables[$variant]->replace( $text );
 490                 }
 491                 return $text;
 492         }
 493
 494         /**
 495          * Call translate() to convert text to all valid variants.
 496          *
 497          * @param string $text The text to be converted
 498          * @return array Variant => converted text
 499          */
 500         public function autoConvertToAllVariants( $text ) {
 501                 $this->loadTables();
 502
 503                 $ret = [];
 504                 foreach ( $this->mVariants as $variant ) {
 505                         $ret[$variant] = $this->translate( $text, $variant );
 506                 }
 507
 508                 return $ret;
 509         }
 510
 511         /**
 512          * Apply manual conversion rules.
 513          *
 514          * @param ConverterRule $convRule
 515          */
 516         protected function applyManualConv( $convRule ) {
 517                 // Use syntax -{T|zh-cn:TitleCN; zh-tw:TitleTw}- to custom
 518                 // title conversion.
 519                 // T26072: $mConvRuleTitle was overwritten by other manual
 520                 // rule(s) not for title, this breaks the title conversion.
 521                 $newConvRuleTitle = $convRule->getTitle();
 522                 if ( $newConvRuleTitle ) {
 523                         // So I add an empty check for getTitle()
 524                         $this->mConvRuleTitle = $newConvRuleTitle;
 525                 }
 526
 527                 // merge/remove manual conversion rules to/from global table
 528                 $convTable = $convRule->getConvTable();
 529                 $action = $convRule->getRulesAction();
 530                 foreach ( $convTable as $variant => $pair ) {
 531                         if ( !$this->validateVariant( $variant ) ) {
 532                                 continue;
 533                         }
 534
 535                         if ( $action == 'add' ) {
 536                                 // More efficient than array_merge(), about 2.5 times.
 537                                 foreach ( $pair as $from => $to ) {
 538                                         $this->mTables[$variant]->setPair( $from, $to );
 539                                 }
 540                         } elseif ( $action == 'remove' ) {
 541                                 $this->mTables[$variant]->removeArray( $pair );
 542                         }
 543                 }
 544         }
 545
 546         /**
 547          * Auto convert a Title object to a readable string in the
 548          * preferred variant.
 549          *
 550          * @param Title $title A object of Title
 551          * @return string Converted title text
 552          */
 553         public function convertTitle( $title ) {
 554                 $variant = $this->getPreferredVariant();
 555                 $index = $title->getNamespace();
 556                 if ( $index !== NS_MAIN ) {
 557                         $text = $this->convertNamespace( $index, $variant ) . ':';
 558                 } else {
 559                         $text = '';
 560                 }
 561                 $text .= $this->translate( $title->getText(), $variant );
 562                 return $text;
 563         }
 564
 565         /**
 566          * Get the namespace display name in the preferred variant.
 567          *
 568          * @param int $index Namespace id
 569          * @param string|null $variant Variant code or null for preferred variant
 570          * @return string Namespace name for display
 571          */
 572         public function convertNamespace( $index, $variant = null ) {
 573                 if ( $index === NS_MAIN ) {
 574                         return '';
 575                 }
 576
 577                 if ( $variant === null ) {
 578                         $variant = $this->getPreferredVariant();
 579                 }
 580
 581                 $cache = MediaWikiServices::getInstance()->getLocalServerObjectCache();
 582                 $key = $cache->makeKey( 'languageconverter', 'namespace-text', $index, $variant );
 583                 $nsVariantText = $cache->get( $key );
 584                 if ( $nsVariantText !== false ) {
 585                         return $nsVariantText;
 586                 }
 587
 588                 // First check if a message gives a converted name in the target variant.
 589                 $nsConvMsg = wfMessage( 'conversion-ns' . $index )->inLanguage( $variant );
 590                 if ( $nsConvMsg->exists() ) {
 591                         $nsVariantText = $nsConvMsg->plain();
 592                 }
 593
 594                 // Then check if a message gives a converted name in content language
 595                 // which needs extra translation to the target variant.
 596                 if ( $nsVariantText === false ) {
 597                         $nsConvMsg = wfMessage( 'conversion-ns' . $index )->inContentLanguage();
 598                         if ( $nsConvMsg->exists() ) {
 599                                 $nsVariantText = $this->translate( $nsConvMsg->plain(), $variant );
 600                         }
 601                 }
 602
 603                 if ( $nsVariantText === false ) {
 604                         // No message exists, retrieve it from the target variant's namespace names.
 605                         $langObj = $this->mLangObj->factory( $variant );
 606                         $nsVariantText = $langObj->getFormattedNsText( $index );
 607                 }
 608
 609                 $cache->set( $key, $nsVariantText, 60 );
 610
 611                 return $nsVariantText;
 612         }
 613
 614         /**
 615          * Convert text to different variants of a language. The automatic
 616          * conversion is done in autoConvert(). Here we parse the text
 617          * marked with -{}-, which specifies special conversions of the
 618          * text that can not be accomplished in autoConvert().
 619          *
 620          * Syntax of the markup:
 621          * -{code1:text1;code2:text2;...}-  or
 622          * -{flags|code1:text1;code2:text2;...}-  or
 623          * -{text}- in which case no conversion should take place for text
 624          *
 625          * @param string $text Text to be converted
 626          * @return string Converted text
 627          */
 628         public function convert( $text ) {
 629                 $variant = $this->getPreferredVariant();
 630                 return $this->convertTo( $text, $variant );
 631         }
 632
 633         /**
 634          * Same as convert() except a extra parameter to custom variant.
 635          *
 636          * @param string $text Text to be converted
 637          * @param string $variant The target variant code
 638          * @return string Converted text
 639          */
 640         public function convertTo( $text, $variant ) {
 641                 global $wgDisableLangConversion;
 642                 if ( $wgDisableLangConversion ) {
 643                         return $text;
 644                 }
 645                 // Reset converter state for a new converter run.
 646                 $this->mConvRuleTitle = false;
 647                 return $this->recursiveConvertTopLevel( $text, $variant );
 648         }
 649
 650         /**
 651          * Recursively convert text on the outside. Allow to use nested
 652          * markups to custom rules.
 653          *
 654          * @param string $text Text to be converted
 655          * @param string $variant The target variant code
 656          * @param int $depth Depth of recursion
 657          * @return string Converted text
 658          */
 659         protected function recursiveConvertTopLevel( $text, $variant, $depth = 0 ) {
 660                 $startPos = 0;
 661                 $out = '';
 662                 $length = strlen( $text );
 663                 $shouldConvert = !$this->guessVariant( $text, $variant );
 664                 $continue = 1;
 665
 666                 $noScript = '<script.*?>.*?<\/script>(*SKIP)(*FAIL)';
 667                 $noStyle = '<style.*?>.*?<\/style>(*SKIP)(*FAIL)';
 668                 // @codingStandardsIgnoreStart Generic.Files.LineLength.TooLong
 669                 $noHtml = '<(?:[^>=]*+(?>[^>=]*+=\s*+(?:"[^"]*"|\'[^\']*\'|[^\'">\s]*+))*+[^>=]*+>|.*+)(*SKIP)(*FAIL)';
 670                 // @codingStandardsIgnoreEnd
 671                 while ( $startPos < $length && $continue ) {
 672                         $continue = preg_match(
 673                                 // Only match -{ outside of html.
 674                                 "/$noScript|$noStyle|$noHtml|-\{/",
 675                                 $text,
 676                                 $m,
 677                                 PREG_OFFSET_CAPTURE,
 678                                 $startPos
 679                         );
 680
 681                         if ( !$continue ) {
 682                                 // No more markup, append final segment
 683                                 $fragment = substr( $text, $startPos );
 684                                 $out .= $shouldConvert ? $this->autoConvert( $fragment, $variant ) : $fragment;
 685                                 return $out;
 686                         }
 687
 688                         // Offset of the match of the regex pattern.
 689                         $pos = $m[0][1];
 690
 691                         // Append initial segment
 692                         $fragment = substr( $text, $startPos, $pos - $startPos );
 693                         $out .= $shouldConvert ? $this->autoConvert( $fragment, $variant ) : $fragment;
 694                         // -{ marker found, not in attribute
 695                         // Advance position up to -{ marker.
 696                         $startPos = $pos;
 697                         // Do recursive conversion
 698                         // Note: This passes $startPos by reference, and advances it.
 699                         $out .= $this->recursiveConvertRule( $text, $variant, $startPos, $depth + 1 );
 700                 }
 701                 return $out;
 702         }
 703
 704         /**
 705          * Recursively convert text on the inside.
 706          *
 707          * @param string $text Text to be converted
 708          * @param string $variant The target variant code
 709          * @param int &$startPos
 710          * @param int $depth Depth of recursion
 711          *
 712          * @throws MWException
 713          * @return string Converted text
 714          */
 715         protected function recursiveConvertRule( $text, $variant, &$startPos, $depth = 0 ) {
 716                 // Quick sanity check (no function calls)
 717                 if ( $text[$startPos] !== '-' || $text[$startPos + 1] !== '{' ) {
 718                         throw new MWException( __METHOD__ . ': invalid input string' );
 719                 }
 720
 721                 $startPos += 2;
 722                 $inner = '';
 723                 $warningDone = false;
 724                 $length = strlen( $text );
 725
 726                 while ( $startPos < $length ) {
 727                         $m = false;
 728                         preg_match( '/-\{|\}-/', $text, $m, PREG_OFFSET_CAPTURE, $startPos );
 729                         if ( !$m ) {
 730                                 // Unclosed rule
 731                                 break;
 732                         }
 733
 734                         $token = $m[0][0];
 735                         $pos = $m[0][1];
 736
 737                         // Markup found
 738                         // Append initial segment
 739                         $inner .= substr( $text, $startPos, $pos - $startPos );
 740
 741                         // Advance position
 742                         $startPos = $pos;
 743
 744                         switch ( $token ) {
 745                                 case '-{':
 746                                         // Check max depth
 747                                         if ( $depth >= $this->mMaxDepth ) {
 748                                                 $inner .= '-{';
 749                                                 if ( !$warningDone ) {
 750                                                         $inner .= '<span class="error">' .
 751                                                                 wfMessage( 'language-converter-depth-warning' )
 752                                                                         ->numParams( $this->mMaxDepth )->inContentLanguage()->text() .
 753                                                                 '</span>';
 754                                                         $warningDone = true;
 755                                                 }
 756                                                 $startPos += 2;
 757                                                 continue;
 758                                         }
 759                                         // Recursively parse another rule
 760                                         $inner .= $this->recursiveConvertRule( $text, $variant, $startPos, $depth + 1 );
 761                                         break;
 762                                 case '}-':
 763                                         // Apply the rule
 764                                         $startPos += 2;
 765                                         $rule = new ConverterRule( $inner, $this );
 766                                         $rule->parse( $variant );
 767                                         $this->applyManualConv( $rule );
 768                                         return $rule->getDisplay();
 769                                 default:
 770                                         throw new MWException( __METHOD__ . ': invalid regex match' );
 771                         }
 772                 }
 773
 774                 // Unclosed rule
 775                 if ( $startPos < $length ) {
 776                         $inner .= substr( $text, $startPos );
 777                 }
 778                 $startPos = $length;
 779                 return '-{' . $this->autoConvert( $inner, $variant );
 780         }
 781
 782         /**
 783          * If a language supports multiple variants, it is possible that
 784          * non-existing link in one variant actually exists in another variant.
 785          * This function tries to find it. See e.g. LanguageZh.php
 786          * The input parameters may be modified upon return
 787          *
 788          * @param string &$link The name of the link
 789          * @param Title &$nt The title object of the link
 790          * @param bool $ignoreOtherCond To disable other conditions when
 791          *   we need to transclude a template or update a category's link
 792          */
 793         public function findVariantLink( &$link, &$nt, $ignoreOtherCond = false ) {
 794                 # If the article has already existed, there is no need to
 795                 # check it again, otherwise it may cause a fault.
 796                 if ( is_object( $nt ) && $nt->exists() ) {
 797                         return;
 798                 }
 799
 800                 global $wgDisableLangConversion, $wgDisableTitleConversion, $wgRequest;
 801                 $isredir = $wgRequest->getText( 'redirect', 'yes' );
 802                 $action = $wgRequest->getText( 'action' );
 803                 if ( $action == 'edit' && $wgRequest->getBool( 'redlink' ) ) {
 804                         $action = 'view';
 805                 }
 806                 $linkconvert = $wgRequest->getText( 'linkconvert', 'yes' );
 807                 $disableLinkConversion = $wgDisableLangConversion
 808                         || $wgDisableTitleConversion;
 809                 $linkBatch = new LinkBatch();
 810
 811                 $ns = NS_MAIN;
 812
 813                 if ( $disableLinkConversion ||
 814                         ( !$ignoreOtherCond &&
 815                                 ( $isredir == 'no'
 816                                         || $action == 'edit'
 817                                         || $action == 'submit'
 818                                         || $linkconvert == 'no' ) ) ) {
 819                         return;
 820                 }
 821
 822                 if ( is_object( $nt ) ) {
 823                         $ns = $nt->getNamespace();
 824                 }
 825
 826                 $variants = $this->autoConvertToAllVariants( $link );
 827                 if ( !$variants ) { // give up
 828                         return;
 829                 }
 830
 831                 $titles = [];
 832
 833                 foreach ( $variants as $v ) {
 834                         if ( $v != $link ) {
 835                                 $varnt = Title::newFromText( $v, $ns );
 836                                 if ( !is_null( $varnt ) ) {
 837                                         $linkBatch->addObj( $varnt );
 838                                         $titles[] = $varnt;
 839                                 }
 840                         }
 841                 }
 842
 843                 // fetch all variants in single query
 844                 $linkBatch->execute();
 845
 846                 foreach ( $titles as $varnt ) {
 847                         if ( $varnt->getArticleID() > 0 ) {
 848                                 $nt = $varnt;
 849                                 $link = $varnt->getText();
 850                                 break;
 851                         }
 852                 }
 853         }
 854
 855         /**
 856          * Returns language specific hash options.
 857          *
 858          * @return string
 859          */
 860         public function getExtraHashOptions() {
 861                 $variant = $this->getPreferredVariant();
 862
 863                 return '!' . $variant;
 864         }
 865
 866         /**
 867          * Guess if a text is written in a variant. This should be implemented in subclasses.
 868          *
 869          * @param string $text The text to be checked
 870          * @param string $variant Language code of the variant to be checked for
 871          * @return bool True if $text appears to be written in $variant, false if not
 872          *
 873          * @author Nikola Smolenski <smolensk@eunet.rs>
 874          * @since 1.19
 875          */
 876         public function guessVariant( $text, $variant ) {
 877                 return false;
 878         }
 879
 880         /**
 881          * Load default conversion tables.
 882          * This method must be implemented in derived class.
 883          *
 884          * @private
 885          * @throws MWException
 886          */
 887         function loadDefaultTables() {
 888                 $class = static::class;
 889                 throw new MWException( "Must implement loadDefaultTables() method in class $class" );
 890         }
 891
 892         /**
 893          * Load conversion tables either from the cache or the disk.
 894          * @private
 895          * @param bool $fromCache Load from memcached? Defaults to true.
 896          */
 897         function loadTables( $fromCache = true ) {
 898                 global $wgLanguageConverterCacheType;
 899
 900                 if ( $this->mTablesLoaded ) {
 901                         return;
 902                 }
 903
 904                 $this->mTablesLoaded = true;
 905                 $this->mTables = false;
 906                 $cache = ObjectCache::getInstance( $wgLanguageConverterCacheType );
 907                 $cacheKey = $cache->makeKey( 'conversiontables', $this->mMainLanguageCode );
 908                 if ( $fromCache ) {
 909                         $this->mTables = $cache->get( $cacheKey );
 910                 }
 911                 if ( !$this->mTables || !array_key_exists( self::CACHE_VERSION_KEY, $this->mTables ) ) {
 912                         // not in cache, or we need a fresh reload.
 913                         // We will first load the default tables
 914                         // then update them using things in MediaWiki:Conversiontable/*
 915                         $this->loadDefaultTables();
 916                         foreach ( $this->mVariants as $var ) {
 917                                 $cached = $this->parseCachedTable( $var );
 918                                 $this->mTables[$var]->mergeArray( $cached );
 919                         }
 920
 921                         $this->postLoadTables();
 922                         $this->mTables[self::CACHE_VERSION_KEY] = true;
 923
 924                         $cache->set( $cacheKey, $this->mTables, 43200 );
 925                 }
 926         }
 927
 928         /**
 929          * Hook for post processing after conversion tables are loaded.
 930          */
 931         function postLoadTables() {
 932         }
 933
 934         /**
 935          * Reload the conversion tables.
 936          *
 937          * Also used by test suites which need to reset the converter state.
 938          *
 939          * @private
 940          */
 941         private function reloadTables() {
 942                 if ( $this->mTables ) {
 943                         unset( $this->mTables );
 944                 }
 945
 946                 $this->mTablesLoaded = false;
 947                 $this->loadTables( false );
 948         }
 949
 950         /**
 951          * Parse the conversion table stored in the cache.
 952          *
 953          * The tables should be in blocks of the following form:
 954          *              -{
 955          *                      word => word ;
 956          *                      word => word ;
 957          *                      ...
 958          *              }-
 959          *
 960          * To make the tables more manageable, subpages are allowed
 961          * and will be parsed recursively if $recursive == true.
 962          *
 963          * @param string $code Language code
 964          * @param string $subpage Subpage name
 965          * @param bool $recursive Parse subpages recursively? Defaults to true.
 966          *
 967          * @return array
 968          */
 969         function parseCachedTable( $code, $subpage = '', $recursive = true ) {
 970                 static $parsed = [];
 971
 972                 $key = 'Conversiontable/' . $code;
 973                 if ( $subpage ) {
 974                         $key .= '/' . $subpage;
 975                 }
 976                 if ( array_key_exists( $key, $parsed ) ) {
 977                         return [];
 978                 }
 979
 980                 $parsed[$key] = true;
 981
 982                 if ( $subpage === '' ) {
 983                         $txt = MessageCache::singleton()->getMsgFromNamespace( $key, $code );
 984                 } else {
 985                         $txt = false;
 986                         $title = Title::makeTitleSafe( NS_MEDIAWIKI, $key );
 987                         if ( $title && $title->exists() ) {
 988                                 $revision = Revision::newFromTitle( $title );
 989                                 if ( $revision ) {
 990                                         if ( $revision->getContentModel() == CONTENT_MODEL_WIKITEXT ) {
 991                                                 $txt = $revision->getContent( Revision::RAW )->getNativeData();
 992                                         }
 993
 994                                         // @todo in the future, use a specialized content model, perhaps based on json!
 995                                 }
 996                         }
 997                 }
 998
 999                 # Nothing to parse if there's no text
1000                 if ( $txt === false || $txt === null || $txt === '' ) {
1001                         return [];
1002                 }
1003
1004                 // get all subpage links of the form
1005                 // [[MediaWiki:Conversiontable/zh-xx/...|...]]
1006                 $linkhead = $this->mLangObj->getNsText( NS_MEDIAWIKI ) .
1007                         ':Conversiontable';
1008                 $subs = StringUtils::explode( '[[', $txt );
1009                 $sublinks = [];
1010                 foreach ( $subs as $sub ) {
1011                         $link = explode( ']]', $sub, 2 );
1012                         if ( count( $link ) != 2 ) {
1013                                 continue;
1014                         }
1015                         $b = explode( '|', $link[0], 2 );
1016                         $b = explode( '/', trim( $b[0] ), 3 );
1017                         if ( count( $b ) == 3 ) {
1018                                 $sublink = $b[2];
1019                         } else {
1020                                 $sublink = '';
1021                         }
1022
1023                         if ( $b[0] == $linkhead && $b[1] == $code ) {
1024                                 $sublinks[] = $sublink;
1025                         }
1026                 }
1027
1028                 // parse the mappings in this page
1029                 $blocks = StringUtils::explode( '-{', $txt );
1030                 $ret = [];
1031                 $first = true;
1032                 foreach ( $blocks as $block ) {
1033                         if ( $first ) {
1034                                 // Skip the part before the first -{
1035                                 $first = false;
1036                                 continue;
1037                         }
1038                         $mappings = explode( '}-', $block, 2 )[0];
1039                         $stripped = str_replace( [ "'", '"', '*', '#' ], '', $mappings );
1040                         $table = StringUtils::explode( ';', $stripped );
1041                         foreach ( $table as $t ) {
1042                                 $m = explode( '=>', $t, 3 );
1043                                 if ( count( $m ) != 2 ) {
1044                                         continue;
1045                                 }
1046                                 // trim any trailling comments starting with '//'
1047                                 $tt = explode( '//', $m[1], 2 );
1048                                 $ret[trim( $m[0] )] = trim( $tt[0] );
1049                         }
1050                 }
1051
1052                 // recursively parse the subpages
1053                 if ( $recursive ) {
1054                         foreach ( $sublinks as $link ) {
1055                                 $s = $this->parseCachedTable( $code, $link, $recursive );
1056                                 $ret = $s + $ret;
1057                         }
1058                 }
1059
1060                 if ( $this->mUcfirst ) {
1061                         foreach ( $ret as $k => $v ) {
1062                                 $ret[$this->mLangObj->ucfirst( $k )] = $this->mLangObj->ucfirst( $v );
1063                         }
1064                 }
1065                 return $ret;
1066         }
1067
1068         /**
1069          * Enclose a string with the "no conversion" tag. This is used by
1070          * various functions in the Parser.
1071          *
1072          * @param string $text Text to be tagged for no conversion
1073          * @param bool $noParse Unused
1074          * @return string The tagged text
1075          */
1076         public function markNoConversion( $text, $noParse = false ) {
1077                 # don't mark if already marked
1078                 if ( strpos( $text, '-{' ) || strpos( $text, '}-' ) ) {
1079                         return $text;
1080                 }
1081
1082                 $ret = "-{R|$text}-";
1083                 return $ret;
1084         }
1085
1086         /**
1087          * Convert the sorting key for category links. This should make different
1088          * keys that are variants of each other map to the same key.
1089          *
1090          * @param string $key
1091          *
1092          * @return string
1093          */
1094         function convertCategoryKey( $key ) {
1095                 return $key;
1096         }
1097
1098         /**
1099          * Refresh the cache of conversion tables when
1100          * MediaWiki:Conversiontable* is updated.
1101          *
1102          * @param Title $titleobj The Title of the page being updated
1103          */
1104         public function updateConversionTable( Title $titleobj ) {
1105                 if ( $titleobj->getNamespace() == NS_MEDIAWIKI ) {
1106                         $title = $titleobj->getDBkey();
1107                         $t = explode( '/', $title, 3 );
1108                         $c = count( $t );
1109                         if ( $c > 1 && $t[0] == 'Conversiontable' ) {
1110                                 if ( $this->validateVariant( $t[1] ) ) {
1111                                         $this->reloadTables();
1112                                 }
1113                         }
1114                 }
1115         }
1116
1117         /**
1118          * Get the cached separator pattern for ConverterRule::parseRules()
1119          * @return string
1120          */
1121         function getVarSeparatorPattern() {
1122                 if ( is_null( $this->mVarSeparatorPattern ) ) {
1123                         // varsep_pattern for preg_split:
1124                         // text should be splited by ";" only if a valid variant
1125                         // name exist after the markup, for example:
1126                         //  -{zh-hans:<span style="font-size:120%;">xxx</span>;zh-hant:\
1127                         //  <span style="font-size:120%;">yyy</span>;}-
1128                         // we should split it as:
1129                         //  [
1130                         //    [0] => 'zh-hans:<span style="font-size:120%;">xxx</span>'
1131                         //    [1] => 'zh-hant:<span style="font-size:120%;">yyy</span>'
1132                         //    [2] => ''
1133                         //  ]
1134                         $pat = '/;\s*(?=';
1135                         foreach ( $this->mVariants as $variant ) {
1136                                 // zh-hans:xxx;zh-hant:yyy
1137                                 $pat .= $variant . '\s*:|';
1138                                 // xxx=>zh-hans:yyy; xxx=>zh-hant:zzz
1139                                 $pat .= '[^;]*?=>\s*' . $variant . '\s*:|';
1140                         }
1141                         $pat .= '\s*$)/';
1142                         $this->mVarSeparatorPattern = $pat;
1143                 }
1144                 return $this->mVarSeparatorPattern;
1145         }
1146 }