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