languages/LanguageConverter.php

   1 <?php
   2 /**
   3  * This program is free software; you can redistribute it and/or modify
   4  * it under the terms of the GNU General Public License as published by
   5  * the Free Software Foundation; either version 2 of the License, or
   6  * (at your option) any later version.
   7  *
   8  * This program is distributed in the hope that it will be useful,
   9  * but WITHOUT ANY WARRANTY; without even the implied warranty of
  10  * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the
  11  * GNU General Public License for more details.
  12  *
  13  * You should have received a copy of the GNU General Public License along
  14  * with this program; if not, write to the Free Software Foundation, Inc.,
  15  * 51 Franklin Street, Fifth Floor, Boston, MA 02110-1301, USA.
  16  * http://www.gnu.org/copyleft/gpl.html
  17  *
  18  * @file
  19  * @ingroup Language
  20  */
  21 use MediaWiki\MediaWikiServices;
  22
  23 use MediaWiki\Logger\LoggerFactory;
  24
  25 /**
  26  * Base class for language conversion.
  27  * @ingroup Language
  28  *
  29  * @author Zhengzhu Feng <zhengzhu@gmail.com>
  30  * @author fdcn <fdcn64@gmail.com>
  31  * @author shinjiman <shinjiman@gmail.com>
  32  * @author PhiLiP <philip.npc@gmail.com>
  33  */
  34 class LanguageConverter {
  35         /**
  36          * languages supporting variants
  37          * @since 1.20
  38          * @var array
  39          */
  40         public static $languagesWithVariants = [
  41                 'en',
  42                 'crh',
  43                 'gan',
  44                 'iu',
  45                 'kk',
  46                 'ku',
  47                 'shi',
  48                 'sr',
  49                 'tg',
  50                 'uz',
  51                 'zh',
  52         ];
  53
  54         public $mMainLanguageCode;
  55
  56         /**
  57          * @var string[]
  58          */
  59         public $mVariants;
  60         public $mVariantFallbacks;
  61         public $mVariantNames;
  62         public $mTablesLoaded = false;
  63         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                 return $this->mVariantFallbacks[$variant] ?? $this->mMainLanguageCode;
 144         }
 145
 146         /**
 147          * Get the title produced by the conversion rule.
 148          * @return string The converted title text
 149          */
 150         public function getConvRuleTitle() {
 151                 return $this->mConvRuleTitle;
 152         }
 153
 154         /**
 155          * Get preferred language variant.
 156          * @return string The preferred language code
 157          */
 158         public function getPreferredVariant() {
 159                 global $wgDefaultLanguageVariant, $wgUser;
 160
 161                 $req = $this->getURLVariant();
 162
 163                 Hooks::run( 'GetLangPreferredVariant', [ &$req ] );
 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                 $req = $this->validateVariant( $req );
 176
 177                 // This function, unlike the other get*Variant functions, is
 178                 // not memoized (i.e. there return value is not cached) since
 179                 // new information might appear during processing after this
 180                 // is first called.
 181                 if ( $req ) {
 182                         return $req;
 183                 }
 184                 return $this->mMainLanguageCode;
 185         }
 186
 187         /**
 188          * Get default variant.
 189          * This function would not be affected by user's settings
 190          * @return string The default variant code
 191          */
 192         public function getDefaultVariant() {
 193                 global $wgDefaultLanguageVariant;
 194
 195                 $req = $this->getURLVariant();
 196
 197                 if ( !$req ) {
 198                         $req = $this->getHeaderVariant();
 199                 }
 200
 201                 if ( $wgDefaultLanguageVariant && !$req ) {
 202                         $req = $this->validateVariant( $wgDefaultLanguageVariant );
 203                 }
 204
 205                 if ( $req ) {
 206                         return $req;
 207                 }
 208                 return $this->mMainLanguageCode;
 209         }
 210
 211         /**
 212          * Validate the variant and return an appropriate strict internal
 213          * variant code if one exists.  Compare to Language::hasVariant()
 214          * which does a strict test.
 215          *
 216          * @param string|null $variant The variant to validate
 217          * @return mixed Returns an equivalent valid variant code if possible,
 218          *   null otherwise
 219          */
 220         public function validateVariant( $variant = null ) {
 221                 if ( $variant === null ) {
 222                         return null;
 223                 }
 224                 // Our internal variants are always lower-case; the variant we
 225                 // are validating may have mixed case.
 226                 $variant = LanguageCode::replaceDeprecatedCodes( strtolower( $variant ) );
 227                 if ( in_array( $variant, $this->mVariants ) ) {
 228                         return $variant;
 229                 }
 230                 // Browsers are supposed to use BCP 47 standard in the
 231                 // Accept-Language header, but not all of our internal
 232                 // mediawiki variant codes are BCP 47.  Map BCP 47 code
 233                 // to our internal code.
 234                 foreach ( $this->mVariants as $v ) {
 235                         // Case-insensitive match (BCP 47 is mixed case)
 236                         if ( strtolower( LanguageCode::bcp47( $v ) ) === $variant ) {
 237                                 return $v;
 238                         }
 239                 }
 240                 return null;
 241         }
 242
 243         /**
 244          * Get the variant specified in the URL
 245          *
 246          * @return mixed Variant if one found, null otherwise
 247          */
 248         public function getURLVariant() {
 249                 global $wgRequest;
 250
 251                 if ( $this->mURLVariant ) {
 252                         return $this->mURLVariant;
 253                 }
 254
 255                 // see if the preference is set in the request
 256                 $ret = $wgRequest->getText( 'variant' );
 257
 258                 if ( !$ret ) {
 259                         $ret = $wgRequest->getVal( 'uselang' );
 260                 }
 261
 262                 $this->mURLVariant = $this->validateVariant( $ret );
 263                 return $this->mURLVariant;
 264         }
 265
 266         /**
 267          * Determine if the user has a variant set.
 268          *
 269          * @return mixed Variant if one found, null otherwise
 270          */
 271         protected function getUserVariant() {
 272                 global $wgUser;
 273
 274                 // memoizing this function wreaks havoc on parserTest.php
 275                 /*
 276                 if ( $this->mUserVariant ) {
 277                         return $this->mUserVariant;
 278                 }
 279                 */
 280
 281                 // Get language variant preference from logged in users
 282                 // Don't call this on stub objects because that causes infinite
 283                 // recursion during initialisation
 284                 if ( !$wgUser->isSafeToLoad() ) {
 285                         return false;
 286                 }
 287                 if ( $wgUser->isLoggedIn() ) {
 288                         if (
 289                                 $this->mMainLanguageCode ==
 290                                 MediaWikiServices::getInstance()->getContentLanguage()->getCode()
 291                         ) {
 292                                 $ret = $wgUser->getOption( 'variant' );
 293                         } else {
 294                                 $ret = $wgUser->getOption( 'variant-' . $this->mMainLanguageCode );
 295                         }
 296                 } else {
 297                         // figure out user lang without constructing wgLang to avoid
 298                         // infinite recursion
 299                         $ret = $wgUser->getOption( 'language' );
 300                 }
 301
 302                 $this->mUserVariant = $this->validateVariant( $ret );
 303                 return $this->mUserVariant;
 304         }
 305
 306         /**
 307          * Determine the language variant from the Accept-Language header.
 308          *
 309          * @return mixed Variant if one found, null otherwise
 310          */
 311         protected function getHeaderVariant() {
 312                 global $wgRequest;
 313
 314                 if ( $this->mHeaderVariant ) {
 315                         return $this->mHeaderVariant;
 316                 }
 317
 318                 // See if some supported language variant is set in the
 319                 // HTTP header.
 320                 $languages = array_keys( $wgRequest->getAcceptLang() );
 321                 if ( empty( $languages ) ) {
 322                         return null;
 323                 }
 324
 325                 $fallbackLanguages = [];
 326                 foreach ( $languages as $language ) {
 327                         $this->mHeaderVariant = $this->validateVariant( $language );
 328                         if ( $this->mHeaderVariant ) {
 329                                 break;
 330                         }
 331
 332                         // To see if there are fallbacks of current language.
 333                         // We record these fallback variants, and process
 334                         // them later.
 335                         $fallbacks = $this->getVariantFallbacks( $language );
 336                         if ( is_string( $fallbacks ) && $fallbacks !== $this->mMainLanguageCode ) {
 337                                 $fallbackLanguages[] = $fallbacks;
 338                         } elseif ( is_array( $fallbacks ) ) {
 339                                 $fallbackLanguages =
 340                                         array_merge( $fallbackLanguages, $fallbacks );
 341                         }
 342                 }
 343
 344                 if ( !$this->mHeaderVariant ) {
 345                         // process fallback languages now
 346                         $fallback_languages = array_unique( $fallbackLanguages );
 347                         foreach ( $fallback_languages as $language ) {
 348                                 $this->mHeaderVariant = $this->validateVariant( $language );
 349                                 if ( $this->mHeaderVariant ) {
 350                                         break;
 351                                 }
 352                         }
 353                 }
 354
 355                 return $this->mHeaderVariant;
 356         }
 357
 358         /**
 359          * Dictionary-based conversion.
 360          * This function would not parse the conversion rules.
 361          * If you want to parse rules, try to use convert() or
 362          * convertTo().
 363          *
 364          * @param string $text The text to be converted
 365          * @param bool|string $toVariant The target language code
 366          * @return string The converted text
 367          */
 368         public function autoConvert( $text, $toVariant = false ) {
 369                 $this->loadTables();
 370
 371                 if ( !$toVariant ) {
 372                         $toVariant = $this->getPreferredVariant();
 373                         if ( !$toVariant ) {
 374                                 return $text;
 375                         }
 376                 }
 377
 378                 if ( $this->guessVariant( $text, $toVariant ) ) {
 379                         return $text;
 380                 }
 381                 /* we convert everything except:
 382                    1. HTML markups (anything between < and >)
 383                    2. HTML entities
 384                    3. placeholders created by the parser
 385                    IMPORTANT: Beware of failure from pcre.backtrack_limit (T124404).
 386                    Minimize use of backtracking where possible.
 387                 */
 388                 $marker = '|' . Parser::MARKER_PREFIX . '[^\x7f]++\x7f';
 389
 390                 // this one is needed when the text is inside an HTML markup
 391                 $htmlfix = '|<[^>\004]++(?=\004$)|^[^<>]*+>';
 392
 393                 // Optimize for the common case where these tags have
 394                 // few or no children. Thus try and possesively get as much as
 395                 // possible, and only engage in backtracking when we hit a '<'.
 396
 397                 // disable convert to variants between <code> tags
 398                 $codefix = '<code>[^<]*+(?:(?:(?!<\/code>).)[^<]*+)*+<\/code>|';
 399                 // disable conversion of <script> tags
 400                 $scriptfix = '<script[^>]*+>[^<]*+(?:(?:(?!<\/script>).)[^<]*+)*+<\/script>|';
 401                 // disable conversion of <pre> tags
 402                 $prefix = '<pre[^>]*+>[^<]*+(?:(?:(?!<\/pre>).)[^<]*+)*+<\/pre>|';
 403                 // The "|.*+)" at the end, is in case we missed some part of html syntax,
 404                 // we will fail securely (hopefully) by matching the rest of the string.
 405                 $htmlFullTag = '<(?:[^>=]*+(?>[^>=]*+=\s*+(?:"[^"]*"|\'[^\']*\'|[^\'">\s]*+))*+[^>=]*+>|.*+)|';
 406
 407                 $reg = '/' . $codefix . $scriptfix . $prefix . $htmlFullTag .
 408                         '&[a-zA-Z#][a-z0-9]++;' . $marker . $htmlfix . '|\004$/s';
 409                 $startPos = 0;
 410                 $sourceBlob = '';
 411                 $literalBlob = '';
 412
 413                 // Guard against delimiter nulls in the input
 414                 // (should never happen: see T159174)
 415                 $text = str_replace( "\000", '', $text );
 416                 $text = str_replace( "\004", '', $text );
 417
 418                 $markupMatches = null;
 419                 $elementMatches = null;
 420
 421                 // We add a marker (\004) at the end of text, to ensure we always match the
 422                 // entire text (Otherwise, pcre.backtrack_limit might cause silent failure)
 423                 while ( $startPos < strlen( $text ) ) {
 424                         if ( preg_match( $reg, $text . "\004", $markupMatches, PREG_OFFSET_CAPTURE, $startPos ) ) {
 425                                 $elementPos = $markupMatches[0][1];
 426                                 $element = $markupMatches[0][0];
 427                                 if ( $element === "\004" ) {
 428                                         // We hit the end.
 429                                         $elementPos = strlen( $text );
 430                                         $element = '';
 431                                 } elseif ( substr( $element, -1 ) === "\004" ) {
 432                                         // This can sometimes happen if we have
 433                                         // unclosed html tags (For example
 434                                         // when converting a title attribute
 435                                         // during a recursive call that contains
 436                                         // a &lt; e.g. <div title="&lt;">.
 437                                         $element = substr( $element, 0, -1 );
 438                                 }
 439                         } else {
 440                                 // If we hit here, then Language Converter could be tricked
 441                                 // into doing an XSS, so we refuse to translate.
 442                                 // If non-crazy input manages to reach this code path,
 443                                 // we should consider it a bug.
 444                                 $log = LoggerFactory::getInstance( 'languageconverter' );
 445                                 $log->error( "Hit pcre.backtrack_limit in " . __METHOD__
 446                                         . ". Disabling language conversion for this page.",
 447                                         [
 448                                                 "method" => __METHOD__,
 449                                                 "variant" => $toVariant,
 450                                                 "startOfText" => substr( $text, 0, 500 )
 451                                         ]
 452                                 );
 453                                 return $text;
 454                         }
 455                         // Queue the part before the markup for translation in a batch
 456                         $sourceBlob .= substr( $text, $startPos, $elementPos - $startPos ) . "\000";
 457
 458                         // Advance to the next position
 459                         $startPos = $elementPos + strlen( $element );
 460
 461                         // Translate any alt or title attributes inside the matched element
 462                         if ( $element !== ''
 463                                 && preg_match( '/^(<[^>\s]*+)\s([^>]*+)(.*+)$/', $element, $elementMatches )
 464                         ) {
 465                                 // FIXME, this decodes entities, so if you have something
 466                                 // like <div title="foo&lt;bar"> the bar won't get
 467                                 // translated since after entity decoding it looks like
 468                                 // unclosed html and we call this method recursively
 469                                 // on attributes.
 470                                 $attrs = Sanitizer::decodeTagAttributes( $elementMatches[2] );
 471                                 // Ensure self-closing tags stay self-closing.
 472                                 $close = substr( $elementMatches[2], -1 ) === '/' ? ' /' : '';
 473                                 $changed = false;
 474                                 foreach ( [ 'title', 'alt' ] as $attrName ) {
 475                                         if ( !isset( $attrs[$attrName] ) ) {
 476                                                 continue;
 477                                         }
 478                                         $attr = $attrs[$attrName];
 479                                         // Don't convert URLs
 480                                         if ( !strpos( $attr, '://' ) ) {
 481                                                 $attr = $this->recursiveConvertTopLevel( $attr, $toVariant );
 482                                         }
 483
 484                                         if ( $attr !== $attrs[$attrName] ) {
 485                                                 $attrs[$attrName] = $attr;
 486                                                 $changed = true;
 487                                         }
 488                                 }
 489                                 if ( $changed ) {
 490                                         $element = $elementMatches[1] . Html::expandAttributes( $attrs ) .
 491                                                 $close . $elementMatches[3];
 492                                 }
 493                         }
 494                         $literalBlob .= $element . "\000";
 495                 }
 496
 497                 // Do the main translation batch
 498                 $translatedBlob = $this->translate( $sourceBlob, $toVariant );
 499
 500                 // Put the output back together
 501                 $translatedIter = StringUtils::explode( "\000", $translatedBlob );
 502                 $literalIter = StringUtils::explode( "\000", $literalBlob );
 503                 $output = '';
 504                 while ( $translatedIter->valid() && $literalIter->valid() ) {
 505                         $output .= $translatedIter->current();
 506                         $output .= $literalIter->current();
 507                         $translatedIter->next();
 508                         $literalIter->next();
 509                 }
 510
 511                 return $output;
 512         }
 513
 514         /**
 515          * Translate a string to a variant.
 516          * Doesn't parse rules or do any of that other stuff, for that use
 517          * convert() or convertTo().
 518          *
 519          * @param string $text Text to convert
 520          * @param string $variant Variant language code
 521          * @return string Translated text
 522          */
 523         public function translate( $text, $variant ) {
 524                 // If $text is empty or only includes spaces, do nothing
 525                 // Otherwise translate it
 526                 if ( trim( $text ) ) {
 527                         $this->loadTables();
 528                         $text = $this->mTables[$variant]->replace( $text );
 529                 }
 530                 return $text;
 531         }
 532
 533         /**
 534          * Call translate() to convert text to all valid variants.
 535          *
 536          * @param string $text The text to be converted
 537          * @return array Variant => converted text
 538          */
 539         public function autoConvertToAllVariants( $text ) {
 540                 $this->loadTables();
 541
 542                 $ret = [];
 543                 foreach ( $this->mVariants as $variant ) {
 544                         $ret[$variant] = $this->translate( $text, $variant );
 545                 }
 546
 547                 return $ret;
 548         }
 549
 550         /**
 551          * Apply manual conversion rules.
 552          *
 553          * @param ConverterRule $convRule
 554          */
 555         protected function applyManualConv( $convRule ) {
 556                 // Use syntax -{T|zh-cn:TitleCN; zh-tw:TitleTw}- to custom
 557                 // title conversion.
 558                 // T26072: $mConvRuleTitle was overwritten by other manual
 559                 // rule(s) not for title, this breaks the title conversion.
 560                 $newConvRuleTitle = $convRule->getTitle();
 561                 if ( $newConvRuleTitle ) {
 562                         // So I add an empty check for getTitle()
 563                         $this->mConvRuleTitle = $newConvRuleTitle;
 564                 }
 565
 566                 // merge/remove manual conversion rules to/from global table
 567                 $convTable = $convRule->getConvTable();
 568                 $action = $convRule->getRulesAction();
 569                 foreach ( $convTable as $variant => $pair ) {
 570                         $v = $this->validateVariant( $variant );
 571                         if ( !$v ) {
 572                                 continue;
 573                         }
 574
 575                         if ( $action == 'add' ) {
 576                                 // More efficient than array_merge(), about 2.5 times.
 577                                 foreach ( $pair as $from => $to ) {
 578                                         $this->mTables[$v]->setPair( $from, $to );
 579                                 }
 580                         } elseif ( $action == 'remove' ) {
 581                                 $this->mTables[$v]->removeArray( $pair );
 582                         }
 583                 }
 584         }
 585
 586         /**
 587          * Auto convert a Title object to a readable string in the
 588          * preferred variant.
 589          *
 590          * @param Title $title A object of Title
 591          * @return string Converted title text
 592          */
 593         public function convertTitle( $title ) {
 594                 $variant = $this->getPreferredVariant();
 595                 $index = $title->getNamespace();
 596                 if ( $index !== NS_MAIN ) {
 597                         $text = $this->convertNamespace( $index, $variant ) . ':';
 598                 } else {
 599                         $text = '';
 600                 }
 601                 $text .= $this->translate( $title->getText(), $variant );
 602                 return $text;
 603         }
 604
 605         /**
 606          * Get the namespace display name in the preferred variant.
 607          *
 608          * @param int $index Namespace id
 609          * @param string|null $variant Variant code or null for preferred variant
 610          * @return string Namespace name for display
 611          */
 612         public function convertNamespace( $index, $variant = null ) {
 613                 if ( $index === NS_MAIN ) {
 614                         return '';
 615                 }
 616
 617                 if ( $variant === null ) {
 618                         $variant = $this->getPreferredVariant();
 619                 }
 620
 621                 $cache = MediaWikiServices::getInstance()->getLocalServerObjectCache();
 622                 $key = $cache->makeKey( 'languageconverter', 'namespace-text', $index, $variant );
 623                 $nsVariantText = $cache->get( $key );
 624                 if ( $nsVariantText !== false ) {
 625                         return $nsVariantText;
 626                 }
 627
 628                 // First check if a message gives a converted name in the target variant.
 629                 $nsConvMsg = wfMessage( 'conversion-ns' . $index )->inLanguage( $variant );
 630                 if ( $nsConvMsg->exists() ) {
 631                         $nsVariantText = $nsConvMsg->plain();
 632                 }
 633
 634                 // Then check if a message gives a converted name in content language
 635                 // which needs extra translation to the target variant.
 636                 if ( $nsVariantText === false ) {
 637                         $nsConvMsg = wfMessage( 'conversion-ns' . $index )->inContentLanguage();
 638                         if ( $nsConvMsg->exists() ) {
 639                                 $nsVariantText = $this->translate( $nsConvMsg->plain(), $variant );
 640                         }
 641                 }
 642
 643                 if ( $nsVariantText === false ) {
 644                         // No message exists, retrieve it from the target variant's namespace names.
 645                         $langObj = $this->mLangObj->factory( $variant );
 646                         $nsVariantText = $langObj->getFormattedNsText( $index );
 647                 }
 648
 649                 $cache->set( $key, $nsVariantText, 60 );
 650
 651                 return $nsVariantText;
 652         }
 653
 654         /**
 655          * Convert text to different variants of a language. The automatic
 656          * conversion is done in autoConvert(). Here we parse the text
 657          * marked with -{}-, which specifies special conversions of the
 658          * text that can not be accomplished in autoConvert().
 659          *
 660          * Syntax of the markup:
 661          * -{code1:text1;code2:text2;...}-  or
 662          * -{flags|code1:text1;code2:text2;...}-  or
 663          * -{text}- in which case no conversion should take place for text
 664          *
 665          * @warning Glossary state is maintained between calls. Never feed this
 666          *   method input that hasn't properly been escaped as it may result in
 667          *   an XSS in subsequent calls, even if those subsequent calls properly
 668          *   escape things.
 669          * @param string $text Text to be converted, already html escaped.
 670          * @return string Converted text (html)
 671          */
 672         public function convert( $text ) {
 673                 $variant = $this->getPreferredVariant();
 674                 return $this->convertTo( $text, $variant );
 675         }
 676
 677         /**
 678          * Same as convert() except a extra parameter to custom variant.
 679          *
 680          * @param string $text Text to be converted, already html escaped
 681          * @param-taint $text exec_html
 682          * @param string $variant The target variant code
 683          * @return string Converted text
 684          * @return-taint escaped
 685          */
 686         public function convertTo( $text, $variant ) {
 687                 global $wgDisableLangConversion;
 688                 if ( $wgDisableLangConversion ) {
 689                         return $text;
 690                 }
 691                 // Reset converter state for a new converter run.
 692                 $this->mConvRuleTitle = false;
 693                 return $this->recursiveConvertTopLevel( $text, $variant );
 694         }
 695
 696         /**
 697          * Recursively convert text on the outside. Allow to use nested
 698          * markups to custom rules.
 699          *
 700          * @param string $text Text to be converted
 701          * @param string $variant The target variant code
 702          * @param int $depth Depth of recursion
 703          * @return string Converted text
 704          */
 705         protected function recursiveConvertTopLevel( $text, $variant, $depth = 0 ) {
 706                 $startPos = 0;
 707                 $out = '';
 708                 $length = strlen( $text );
 709                 $shouldConvert = !$this->guessVariant( $text, $variant );
 710                 $continue = 1;
 711
 712                 $noScript = '<script.*?>.*?<\/script>(*SKIP)(*FAIL)';
 713                 $noStyle = '<style.*?>.*?<\/style>(*SKIP)(*FAIL)';
 714                 // phpcs:ignore Generic.Files.LineLength
 715                 $noHtml = '<(?:[^>=]*+(?>[^>=]*+=\s*+(?:"[^"]*"|\'[^\']*\'|[^\'">\s]*+))*+[^>=]*+>|.*+)(*SKIP)(*FAIL)';
 716                 while ( $startPos < $length && $continue ) {
 717                         $continue = preg_match(
 718                                 // Only match -{ outside of html.
 719                                 "/$noScript|$noStyle|$noHtml|-\{/",
 720                                 $text,
 721                                 $m,
 722                                 PREG_OFFSET_CAPTURE,
 723                                 $startPos
 724                         );
 725
 726                         if ( !$continue ) {
 727                                 // No more markup, append final segment
 728                                 $fragment = substr( $text, $startPos );
 729                                 $out .= $shouldConvert ? $this->autoConvert( $fragment, $variant ) : $fragment;
 730                                 return $out;
 731                         }
 732
 733                         // Offset of the match of the regex pattern.
 734                         $pos = $m[0][1];
 735
 736                         // Append initial segment
 737                         $fragment = substr( $text, $startPos, $pos - $startPos );
 738                         $out .= $shouldConvert ? $this->autoConvert( $fragment, $variant ) : $fragment;
 739                         // -{ marker found, not in attribute
 740                         // Advance position up to -{ marker.
 741                         $startPos = $pos;
 742                         // Do recursive conversion
 743                         // Note: This passes $startPos by reference, and advances it.
 744                         $out .= $this->recursiveConvertRule( $text, $variant, $startPos, $depth + 1 );
 745                 }
 746                 return $out;
 747         }
 748
 749         /**
 750          * Recursively convert text on the inside.
 751          *
 752          * @param string $text Text to be converted
 753          * @param string $variant The target variant code
 754          * @param int &$startPos
 755          * @param int $depth Depth of recursion
 756          *
 757          * @throws MWException
 758          * @return string Converted text
 759          */
 760         protected function recursiveConvertRule( $text, $variant, &$startPos, $depth = 0 ) {
 761                 // Quick sanity check (no function calls)
 762                 if ( $text[$startPos] !== '-' || $text[$startPos + 1] !== '{' ) {
 763                         throw new MWException( __METHOD__ . ': invalid input string' );
 764                 }
 765
 766                 $startPos += 2;
 767                 $inner = '';
 768                 $warningDone = false;
 769                 $length = strlen( $text );
 770
 771                 while ( $startPos < $length ) {
 772                         $m = false;
 773                         preg_match( '/-\{|\}-/', $text, $m, PREG_OFFSET_CAPTURE, $startPos );
 774                         if ( !$m ) {
 775                                 // Unclosed rule
 776                                 break;
 777                         }
 778
 779                         $token = $m[0][0];
 780                         $pos = $m[0][1];
 781
 782                         // Markup found
 783                         // Append initial segment
 784                         $inner .= substr( $text, $startPos, $pos - $startPos );
 785
 786                         // Advance position
 787                         $startPos = $pos;
 788
 789                         switch ( $token ) {
 790                                 case '-{':
 791                                         // Check max depth
 792                                         if ( $depth >= $this->mMaxDepth ) {
 793                                                 $inner .= '-{';
 794                                                 if ( !$warningDone ) {
 795                                                         $inner .= '<span class="error">' .
 796                                                                 wfMessage( 'language-converter-depth-warning' )
 797                                                                         ->numParams( $this->mMaxDepth )->inContentLanguage()->text() .
 798                                                                 '</span>';
 799                                                         $warningDone = true;
 800                                                 }
 801                                                 $startPos += 2;
 802                                                 break;
 803                                         }
 804                                         // Recursively parse another rule
 805                                         $inner .= $this->recursiveConvertRule( $text, $variant, $startPos, $depth + 1 );
 806                                         break;
 807                                 case '}-':
 808                                         // Apply the rule
 809                                         $startPos += 2;
 810                                         $rule = new ConverterRule( $inner, $this );
 811                                         $rule->parse( $variant );
 812                                         $this->applyManualConv( $rule );
 813                                         return $rule->getDisplay();
 814                                 default:
 815                                         throw new MWException( __METHOD__ . ': invalid regex match' );
 816                         }
 817                 }
 818
 819                 // Unclosed rule
 820                 if ( $startPos < $length ) {
 821                         $inner .= substr( $text, $startPos );
 822                 }
 823                 $startPos = $length;
 824                 return '-{' . $this->autoConvert( $inner, $variant );
 825         }
 826
 827         /**
 828          * If a language supports multiple variants, it is possible that
 829          * non-existing link in one variant actually exists in another variant.
 830          * This function tries to find it. See e.g. LanguageZh.php
 831          * The input parameters may be modified upon return
 832          *
 833          * @param string &$link The name of the link
 834          * @param Title &$nt The title object of the link
 835          * @param bool $ignoreOtherCond To disable other conditions when
 836          *   we need to transclude a template or update a category's link
 837          */
 838         public function findVariantLink( &$link, &$nt, $ignoreOtherCond = false ) {
 839                 # If the article has already existed, there is no need to
 840                 # check it again, otherwise it may cause a fault.
 841                 if ( is_object( $nt ) && $nt->exists() ) {
 842                         return;
 843                 }
 844
 845                 global $wgDisableLangConversion, $wgDisableTitleConversion, $wgRequest;
 846                 $isredir = $wgRequest->getText( 'redirect', 'yes' );
 847                 $action = $wgRequest->getText( 'action' );
 848                 if ( $action == 'edit' && $wgRequest->getBool( 'redlink' ) ) {
 849                         $action = 'view';
 850                 }
 851                 $linkconvert = $wgRequest->getText( 'linkconvert', 'yes' );
 852                 $disableLinkConversion = $wgDisableLangConversion
 853                         || $wgDisableTitleConversion;
 854                 $linkBatch = new LinkBatch();
 855
 856                 $ns = NS_MAIN;
 857
 858                 if ( $disableLinkConversion ||
 859                         ( !$ignoreOtherCond &&
 860                                 ( $isredir == 'no'
 861                                         || $action == 'edit'
 862                                         || $action == 'submit'
 863                                         || $linkconvert == 'no' ) ) ) {
 864                         return;
 865                 }
 866
 867                 if ( is_object( $nt ) ) {
 868                         $ns = $nt->getNamespace();
 869                 }
 870
 871                 $variants = $this->autoConvertToAllVariants( $link );
 872                 if ( !$variants ) { // give up
 873                         return;
 874                 }
 875
 876                 $titles = [];
 877
 878                 foreach ( $variants as $v ) {
 879                         if ( $v != $link ) {
 880                                 $varnt = Title::newFromText( $v, $ns );
 881                                 if ( !is_null( $varnt ) ) {
 882                                         $linkBatch->addObj( $varnt );
 883                                         $titles[] = $varnt;
 884                                 }
 885                         }
 886                 }
 887
 888                 // fetch all variants in single query
 889                 $linkBatch->execute();
 890
 891                 foreach ( $titles as $varnt ) {
 892                         if ( $varnt->getArticleID() > 0 ) {
 893                                 $nt = $varnt;
 894                                 $link = $varnt->getText();
 895                                 break;
 896                         }
 897                 }
 898         }
 899
 900         /**
 901          * Returns language specific hash options.
 902          *
 903          * @return string
 904          */
 905         public function getExtraHashOptions() {
 906                 $variant = $this->getPreferredVariant();
 907
 908                 return '!' . $variant;
 909         }
 910
 911         /**
 912          * Guess if a text is written in a variant. This should be implemented in subclasses.
 913          *
 914          * @param string $text The text to be checked
 915          * @param string $variant Language code of the variant to be checked for
 916          * @return bool True if $text appears to be written in $variant, false if not
 917          *
 918          * @author Nikola Smolenski <smolensk@eunet.rs>
 919          * @since 1.19
 920          */
 921         public function guessVariant( $text, $variant ) {
 922                 return false;
 923         }
 924
 925         /**
 926          * Load default conversion tables.
 927          * This method must be implemented in derived class.
 928          *
 929          * @private
 930          * @throws MWException
 931          */
 932         function loadDefaultTables() {
 933                 $class = static::class;
 934                 throw new MWException( "Must implement loadDefaultTables() method in class $class" );
 935         }
 936
 937         /**
 938          * Load conversion tables either from the cache or the disk.
 939          * @private
 940          * @param bool $fromCache Load from memcached? Defaults to true.
 941          */
 942         function loadTables( $fromCache = true ) {
 943                 global $wgLanguageConverterCacheType;
 944
 945                 if ( $this->mTablesLoaded ) {
 946                         return;
 947                 }
 948
 949                 $this->mTablesLoaded = true;
 950                 $this->mTables = false;
 951                 $cache = ObjectCache::getInstance( $wgLanguageConverterCacheType );
 952                 $cacheKey = $cache->makeKey( 'conversiontables', $this->mMainLanguageCode );
 953                 if ( $fromCache ) {
 954                         $this->mTables = $cache->get( $cacheKey );
 955                 }
 956                 if ( !$this->mTables || !array_key_exists( self::CACHE_VERSION_KEY, $this->mTables ) ) {
 957                         // not in cache, or we need a fresh reload.
 958                         // We will first load the default tables
 959                         // then update them using things in MediaWiki:Conversiontable/*
 960                         $this->loadDefaultTables();
 961                         foreach ( $this->mVariants as $var ) {
 962                                 $cached = $this->parseCachedTable( $var );
 963                                 $this->mTables[$var]->mergeArray( $cached );
 964                         }
 965
 966                         $this->postLoadTables();
 967                         $this->mTables[self::CACHE_VERSION_KEY] = true;
 968
 969                         $cache->set( $cacheKey, $this->mTables, 43200 );
 970                 }
 971         }
 972
 973         /**
 974          * Hook for post processing after conversion tables are loaded.
 975          */
 976         function postLoadTables() {
 977         }
 978
 979         /**
 980          * Reload the conversion tables.
 981          *
 982          * Also used by test suites which need to reset the converter state.
 983          *
 984          * @private
 985          */
 986         private function reloadTables() {
 987                 if ( $this->mTables ) {
 988                         unset( $this->mTables );
 989                 }
 990
 991                 $this->mTablesLoaded = false;
 992                 $this->loadTables( false );
 993         }
 994
 995         /**
 996          * Parse the conversion table stored in the cache.
 997          *
 998          * The tables should be in blocks of the following form:
 999          *              -{
1000          *                      word => word ;
1001          *                      word => word ;
1002          *                      ...
1003          *              }-
1004          *
1005          * To make the tables more manageable, subpages are allowed
1006          * and will be parsed recursively if $recursive == true.
1007          *
1008          * @param string $code Language code
1009          * @param string $subpage Subpage name
1010          * @param bool $recursive Parse subpages recursively? Defaults to true.
1011          *
1012          * @return array
1013          */
1014         function parseCachedTable( $code, $subpage = '', $recursive = true ) {
1015                 static $parsed = [];
1016
1017                 $key = 'Conversiontable/' . $code;
1018                 if ( $subpage ) {
1019                         $key .= '/' . $subpage;
1020                 }
1021                 if ( array_key_exists( $key, $parsed ) ) {
1022                         return [];
1023                 }
1024
1025                 $parsed[$key] = true;
1026
1027                 if ( $subpage === '' ) {
1028                         $txt = MessageCache::singleton()->getMsgFromNamespace( $key, $code );
1029                 } else {
1030                         $txt = false;
1031                         $title = Title::makeTitleSafe( NS_MEDIAWIKI, $key );
1032                         if ( $title && $title->exists() ) {
1033                                 $revision = Revision::newFromTitle( $title );
1034                                 if ( $revision ) {
1035                                         if ( $revision->getContentModel() == CONTENT_MODEL_WIKITEXT ) {
1036                                                 $txt = $revision->getContent( Revision::RAW )->getText();
1037                                         }
1038
1039                                         // @todo in the future, use a specialized content model, perhaps based on json!
1040                                 }
1041                         }
1042                 }
1043
1044                 # Nothing to parse if there's no text
1045                 if ( $txt === false || $txt === null || $txt === '' ) {
1046                         return [];
1047                 }
1048
1049                 // get all subpage links of the form
1050                 // [[MediaWiki:Conversiontable/zh-xx/...|...]]
1051                 $linkhead = $this->mLangObj->getNsText( NS_MEDIAWIKI ) .
1052                         ':Conversiontable';
1053                 $subs = StringUtils::explode( '[[', $txt );
1054                 $sublinks = [];
1055                 foreach ( $subs as $sub ) {
1056                         $link = explode( ']]', $sub, 2 );
1057                         if ( count( $link ) != 2 ) {
1058                                 continue;
1059                         }
1060                         $b = explode( '|', $link[0], 2 );
1061                         $b = explode( '/', trim( $b[0] ), 3 );
1062                         if ( count( $b ) == 3 ) {
1063                                 $sublink = $b[2];
1064                         } else {
1065                                 $sublink = '';
1066                         }
1067
1068                         if ( $b[0] == $linkhead && $b[1] == $code ) {
1069                                 $sublinks[] = $sublink;
1070                         }
1071                 }
1072
1073                 // parse the mappings in this page
1074                 $blocks = StringUtils::explode( '-{', $txt );
1075                 $ret = [];
1076                 $first = true;
1077                 foreach ( $blocks as $block ) {
1078                         if ( $first ) {
1079                                 // Skip the part before the first -{
1080                                 $first = false;
1081                                 continue;
1082                         }
1083                         $mappings = explode( '}-', $block, 2 )[0];
1084                         $stripped = str_replace( [ "'", '"', '*', '#' ], '', $mappings );
1085                         $table = StringUtils::explode( ';', $stripped );
1086                         foreach ( $table as $t ) {
1087                                 $m = explode( '=>', $t, 3 );
1088                                 if ( count( $m ) != 2 ) {
1089                                         continue;
1090                                 }
1091                                 // trim any trailling comments starting with '//'
1092                                 $tt = explode( '//', $m[1], 2 );
1093                                 $ret[trim( $m[0] )] = trim( $tt[0] );
1094                         }
1095                 }
1096
1097                 // recursively parse the subpages
1098                 if ( $recursive ) {
1099                         foreach ( $sublinks as $link ) {
1100                                 $s = $this->parseCachedTable( $code, $link, $recursive );
1101                                 $ret = $s + $ret;
1102                         }
1103                 }
1104
1105                 if ( $this->mUcfirst ) {
1106                         foreach ( $ret as $k => $v ) {
1107                                 $ret[$this->mLangObj->ucfirst( $k )] = $this->mLangObj->ucfirst( $v );
1108                         }
1109                 }
1110                 return $ret;
1111         }
1112
1113         /**
1114          * Enclose a string with the "no conversion" tag. This is used by
1115          * various functions in the Parser.
1116          *
1117          * @param string $text Text to be tagged for no conversion
1118          * @param bool $noParse Unused
1119          * @return string The tagged text
1120          */
1121         public function markNoConversion( $text, $noParse = false ) {
1122                 # don't mark if already marked
1123                 if ( strpos( $text, '-{' ) || strpos( $text, '}-' ) ) {
1124                         return $text;
1125                 }
1126
1127                 $ret = "-{R|$text}-";
1128                 return $ret;
1129         }
1130
1131         /**
1132          * Convert the sorting key for category links. This should make different
1133          * keys that are variants of each other map to the same key.
1134          *
1135          * @param string $key
1136          *
1137          * @return string
1138          */
1139         function convertCategoryKey( $key ) {
1140                 return $key;
1141         }
1142
1143         /**
1144          * Refresh the cache of conversion tables when
1145          * MediaWiki:Conversiontable* is updated.
1146          *
1147          * @param Title $titleobj The Title of the page being updated
1148          */
1149         public function updateConversionTable( Title $titleobj ) {
1150                 if ( $titleobj->getNamespace() == NS_MEDIAWIKI ) {
1151                         $title = $titleobj->getDBkey();
1152                         $t = explode( '/', $title, 3 );
1153                         $c = count( $t );
1154                         if ( $c > 1 && $t[0] == 'Conversiontable' ) {
1155                                 if ( $this->validateVariant( $t[1] ) ) {
1156                                         $this->reloadTables();
1157                                 }
1158                         }
1159                 }
1160         }
1161
1162         /**
1163          * Get the cached separator pattern for ConverterRule::parseRules()
1164          * @return string
1165          */
1166         function getVarSeparatorPattern() {
1167                 if ( is_null( $this->mVarSeparatorPattern ) ) {
1168                         // varsep_pattern for preg_split:
1169                         // text should be splited by ";" only if a valid variant
1170                         // name exist after the markup, for example:
1171                         //  -{zh-hans:<span style="font-size:120%;">xxx</span>;zh-hant:\
1172                         //  <span style="font-size:120%;">yyy</span>;}-
1173                         // we should split it as:
1174                         //  [
1175                         //    [0] => 'zh-hans:<span style="font-size:120%;">xxx</span>'
1176                         //    [1] => 'zh-hant:<span style="font-size:120%;">yyy</span>'
1177                         //    [2] => ''
1178                         //  ]
1179                         $expandedVariants = [];
1180                         foreach ( $this->mVariants as $variant ) {
1181                                 $expandedVariants[ $variant ] = 1;
1182                                 // Accept standard BCP 47 names for variants as well.
1183                                 $expandedVariants[ LanguageCode::bcp47( $variant ) ] = 1;
1184                         }
1185                         // Accept old deprecated names for variants
1186                         foreach ( LanguageCode::getDeprecatedCodeMapping() as $old => $new ) {
1187                                 if ( isset( $expandedVariants[ $new ] ) ) {
1188                                         $expandedVariants[ $old ] = 1;
1189                                 }
1190                         }
1191
1192                         $pat = '/;\s*(?=';
1193                         foreach ( $expandedVariants as $variant => $ignore ) {
1194                                 // zh-hans:xxx;zh-hant:yyy
1195                                 $pat .= $variant . '\s*:|';
1196                                 // xxx=>zh-hans:yyy; xxx=>zh-hant:zzz
1197                                 $pat .= '[^;]*?=>\s*' . $variant . '\s*:|';
1198                         }
1199                         $pat .= '\s*$)/';
1200                         $this->mVarSeparatorPattern = $pat;
1201                 }
1202                 return $this->mVarSeparatorPattern;
1203         }
1204 }