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