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