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 /**
  24  * Base class for language conversion.
  25  * @ingroup Language
  26  *
  27  * @author Zhengzhu Feng <zhengzhu@gmail.com>
  28  * @author fdcn <fdcn64@gmail.com>
  29  * @author shinjiman <shinjiman@gmail.com>
  30  * @author PhiLiP <philip.npc@gmail.com>
  31  */
  32 class LanguageConverter {
  33         /**
  34          * languages supporting variants
  35          * @since 1.20
  36          * @var array
  37          */
  38         static public $languagesWithVariants = [
  39                 'gan',
  40                 'iu',
  41                 'kk',
  42                 'ku',
  43                 'shi',
  44                 'sr',
  45                 'tg',
  46                 'uz',
  47                 'zh',
  48         ];
  49
  50         public $mMainLanguageCode;
  51         public $mVariants;
  52         public $mVariantFallbacks;
  53         public $mVariantNames;
  54         public $mTablesLoaded = false;
  55         public $mTables;
  56         // 'bidirectional' 'unidirectional' 'disable' for each variant
  57         public $mManualLevel;
  58
  59         /**
  60          * @var string Memcached key name
  61          */
  62         public $mCacheKey;
  63
  64         public $mLangObj;
  65         public $mFlags;
  66         public $mDescCodeSep = ':', $mDescVarSep = ';';
  67         public $mUcfirst = false;
  68         public $mConvRuleTitle = false;
  69         public $mURLVariant;
  70         public $mUserVariant;
  71         public $mHeaderVariant;
  72         public $mMaxDepth = 10;
  73         public $mVarSeparatorPattern;
  74
  75         const CACHE_VERSION_KEY = 'VERSION 7';
  76
  77         /**
  78          * Constructor
  79          *
  80          * @param Language $langobj
  81          * @param string $maincode The main language code of this language
  82          * @param array $variants The supported variants of this language
  83          * @param array $variantfallbacks The fallback language of each variant
  84          * @param array $flags Defining the custom strings that maps to the flags
  85          * @param array $manualLevel Limit for supported variants
  86          */
  87         public function __construct( $langobj, $maincode, $variants = [],
  88                                                                 $variantfallbacks = [], $flags = [],
  89                                                                 $manualLevel = [] ) {
  90                 global $wgDisabledVariants;
  91                 $this->mLangObj = $langobj;
  92                 $this->mMainLanguageCode = $maincode;
  93                 $this->mVariants = array_diff( $variants, $wgDisabledVariants );
  94                 $this->mVariantFallbacks = $variantfallbacks;
  95                 $this->mVariantNames = Language::fetchLanguageNames();
  96                 $this->mCacheKey = wfMemcKey( 'conversiontables', $maincode );
  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 array 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
 346                 $this->loadTables();
 347
 348                 if ( !$toVariant ) {
 349                         $toVariant = $this->getPreferredVariant();
 350                         if ( !$toVariant ) {
 351                                 return $text;
 352                         }
 353                 }
 354
 355                 if ( $this->guessVariant( $text, $toVariant ) ) {
 356                         return $text;
 357                 }
 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                  */
 364                 $marker = '|' . Parser::MARKER_PREFIX . '[\-a-zA-Z0-9]+';
 365
 366                 // this one is needed when the text is inside an HTML markup
 367                 $htmlfix = '|<[^>]+$|^[^<>]*>';
 368
 369                 // disable convert to variants between <code> tags
 370                 $codefix = '<code>.+?<\/code>|';
 371                 // disable conversion of <script> tags
 372                 $scriptfix = '<script.*?>.*?<\/script>|';
 373                 // disable conversion of <pre> tags
 374                 $prefix = '<pre.*?>.*?<\/pre>|';
 375
 376                 $reg = '/' . $codefix . $scriptfix . $prefix .
 377                         '<[^>]+>|&[a-zA-Z#][a-z0-9]+;' . $marker . $htmlfix . '/s';
 378                 $startPos = 0;
 379                 $sourceBlob = '';
 380                 $literalBlob = '';
 381
 382                 // Guard against delimiter nulls in the input
 383                 // (should never happen: see T159174)
 384                 $text = str_replace( "\000", '', $text );
 385
 386                 $markupMatches = null;
 387                 $elementMatches = null;
 388                 while ( $startPos < strlen( $text ) ) {
 389                         if ( preg_match( $reg, $text, $markupMatches, PREG_OFFSET_CAPTURE, $startPos ) ) {
 390                                 $elementPos = $markupMatches[0][1];
 391                                 $element = $markupMatches[0][0];
 392                         } else {
 393                                 $elementPos = strlen( $text );
 394                                 $element = '';
 395                         }
 396
 397                         // Queue the part before the markup for translation in a batch
 398                         $sourceBlob .= substr( $text, $startPos, $elementPos - $startPos ) . "\000";
 399
 400                         // Advance to the next position
 401                         $startPos = $elementPos + strlen( $element );
 402
 403                         // Translate any alt or title attributes inside the matched element
 404                         if ( $element !== ''
 405                                 && preg_match( '/^(<[^>\s]*)\s([^>]*)(.*)$/', $element, $elementMatches )
 406                         ) {
 407                                 $attrs = Sanitizer::decodeTagAttributes( $elementMatches[2] );
 408                                 $changed = false;
 409                                 foreach ( [ 'title', 'alt' ] as $attrName ) {
 410                                         if ( !isset( $attrs[$attrName] ) ) {
 411                                                 continue;
 412                                         }
 413                                         $attr = $attrs[$attrName];
 414                                         // Don't convert URLs
 415                                         if ( !strpos( $attr, '://' ) ) {
 416                                                 $attr = $this->recursiveConvertTopLevel( $attr, $toVariant );
 417                                         }
 418
 419                                         if ( $attr !== $attrs[$attrName] ) {
 420                                                 $attrs[$attrName] = $attr;
 421                                                 $changed = true;
 422                                         }
 423                                 }
 424                                 if ( $changed ) {
 425                                         $element = $elementMatches[1] . Html::expandAttributes( $attrs ) .
 426                                                 $elementMatches[3];
 427                                 }
 428                         }
 429                         $literalBlob .= $element . "\000";
 430                 }
 431
 432                 // Do the main translation batch
 433                 $translatedBlob = $this->translate( $sourceBlob, $toVariant );
 434
 435                 // Put the output back together
 436                 $translatedIter = StringUtils::explode( "\000", $translatedBlob );
 437                 $literalIter = StringUtils::explode( "\000", $literalBlob );
 438                 $output = '';
 439                 while ( $translatedIter->valid() && $literalIter->valid() ) {
 440                         $output .= $translatedIter->current();
 441                         $output .= $literalIter->current();
 442                         $translatedIter->next();
 443                         $literalIter->next();
 444                 }
 445
 446                 return $output;
 447         }
 448
 449         /**
 450          * Translate a string to a variant.
 451          * Doesn't parse rules or do any of that other stuff, for that use
 452          * convert() or convertTo().
 453          *
 454          * @param string $text Text to convert
 455          * @param string $variant Variant language code
 456          * @return string Translated text
 457          */
 458         public function translate( $text, $variant ) {
 459                 // If $text is empty or only includes spaces, do nothing
 460                 // Otherwise translate it
 461                 if ( trim( $text ) ) {
 462                         $this->loadTables();
 463                         $text = $this->mTables[$variant]->replace( $text );
 464                 }
 465                 return $text;
 466         }
 467
 468         /**
 469          * Call translate() to convert text to all valid variants.
 470          *
 471          * @param string $text The text to be converted
 472          * @return array Variant => converted text
 473          */
 474         public function autoConvertToAllVariants( $text ) {
 475                 $this->loadTables();
 476
 477                 $ret = [];
 478                 foreach ( $this->mVariants as $variant ) {
 479                         $ret[$variant] = $this->translate( $text, $variant );
 480                 }
 481
 482                 return $ret;
 483         }
 484
 485         /**
 486          * Apply manual conversion rules.
 487          *
 488          * @param ConverterRule $convRule
 489          */
 490         protected function applyManualConv( $convRule ) {
 491                 // Use syntax -{T|zh-cn:TitleCN; zh-tw:TitleTw}- to custom
 492                 // title conversion.
 493                 // T26072: $mConvRuleTitle was overwritten by other manual
 494                 // rule(s) not for title, this breaks the title conversion.
 495                 $newConvRuleTitle = $convRule->getTitle();
 496                 if ( $newConvRuleTitle ) {
 497                         // So I add an empty check for getTitle()
 498                         $this->mConvRuleTitle = $newConvRuleTitle;
 499                 }
 500
 501                 // merge/remove manual conversion rules to/from global table
 502                 $convTable = $convRule->getConvTable();
 503                 $action = $convRule->getRulesAction();
 504                 foreach ( $convTable as $variant => $pair ) {
 505                         if ( !$this->validateVariant( $variant ) ) {
 506                                 continue;
 507                         }
 508
 509                         if ( $action == 'add' ) {
 510                                 // More efficient than array_merge(), about 2.5 times.
 511                                 foreach ( $pair as $from => $to ) {
 512                                         $this->mTables[$variant]->setPair( $from, $to );
 513                                 }
 514                         } elseif ( $action == 'remove' ) {
 515                                 $this->mTables[$variant]->removeArray( $pair );
 516                         }
 517                 }
 518         }
 519
 520         /**
 521          * Auto convert a Title object to a readable string in the
 522          * preferred variant.
 523          *
 524          * @param Title $title A object of Title
 525          * @return string Converted title text
 526          */
 527         public function convertTitle( $title ) {
 528                 $variant = $this->getPreferredVariant();
 529                 $index = $title->getNamespace();
 530                 if ( $index !== NS_MAIN ) {
 531                         $text = $this->convertNamespace( $index, $variant ) . ':';
 532                 } else {
 533                         $text = '';
 534                 }
 535                 $text .= $this->translate( $title->getText(), $variant );
 536                 return $text;
 537         }
 538
 539         /**
 540          * Get the namespace display name in the preferred variant.
 541          *
 542          * @param int $index Namespace id
 543          * @param string|null $variant Variant code or null for preferred variant
 544          * @return string Namespace name for display
 545          */
 546         public function convertNamespace( $index, $variant = null ) {
 547                 if ( $index === NS_MAIN ) {
 548                         return '';
 549                 }
 550
 551                 if ( $variant === null ) {
 552                         $variant = $this->getPreferredVariant();
 553                 }
 554
 555                 $cache = MediaWikiServices::getInstance()->getLocalServerObjectCache();
 556                 $key = $cache->makeKey( 'languageconverter', 'namespace-text', $index, $variant );
 557                 $nsVariantText = $cache->get( $key );
 558                 if ( $nsVariantText !== false ) {
 559                         return $nsVariantText;
 560                 }
 561
 562                 // First check if a message gives a converted name in the target variant.
 563                 $nsConvMsg = wfMessage( 'conversion-ns' . $index )->inLanguage( $variant );
 564                 if ( $nsConvMsg->exists() ) {
 565                         $nsVariantText = $nsConvMsg->plain();
 566                 }
 567
 568                 // Then check if a message gives a converted name in content language
 569                 // which needs extra translation to the target variant.
 570                 if ( $nsVariantText === false ) {
 571                         $nsConvMsg = wfMessage( 'conversion-ns' . $index )->inContentLanguage();
 572                         if ( $nsConvMsg->exists() ) {
 573                                 $nsVariantText = $this->translate( $nsConvMsg->plain(), $variant );
 574                         }
 575                 }
 576
 577                 if ( $nsVariantText === false ) {
 578                         // No message exists, retrieve it from the target variant's namespace names.
 579                         $langObj = $this->mLangObj->factory( $variant );
 580                         $nsVariantText = $langObj->getFormattedNsText( $index );
 581                 }
 582
 583                 $cache->set( $key, $nsVariantText, 60 );
 584
 585                 return $nsVariantText;
 586         }
 587
 588         /**
 589          * Convert text to different variants of a language. The automatic
 590          * conversion is done in autoConvert(). Here we parse the text
 591          * marked with -{}-, which specifies special conversions of the
 592          * text that can not be accomplished in autoConvert().
 593          *
 594          * Syntax of the markup:
 595          * -{code1:text1;code2:text2;...}-  or
 596          * -{flags|code1:text1;code2:text2;...}-  or
 597          * -{text}- in which case no conversion should take place for text
 598          *
 599          * @param string $text Text to be converted
 600          * @return string Converted text
 601          */
 602         public function convert( $text ) {
 603                 $variant = $this->getPreferredVariant();
 604                 return $this->convertTo( $text, $variant );
 605         }
 606
 607         /**
 608          * Same as convert() except a extra parameter to custom variant.
 609          *
 610          * @param string $text Text to be converted
 611          * @param string $variant The target variant code
 612          * @return string Converted text
 613          */
 614         public function convertTo( $text, $variant ) {
 615                 global $wgDisableLangConversion;
 616                 if ( $wgDisableLangConversion ) {
 617                         return $text;
 618                 }
 619                 // Reset converter state for a new converter run.
 620                 $this->mConvRuleTitle = false;
 621                 return $this->recursiveConvertTopLevel( $text, $variant );
 622         }
 623
 624         /**
 625          * Recursively convert text on the outside. Allow to use nested
 626          * markups to custom rules.
 627          *
 628          * @param string $text Text to be converted
 629          * @param string $variant The target variant code
 630          * @param int $depth Depth of recursion
 631          * @return string Converted text
 632          */
 633         protected function recursiveConvertTopLevel( $text, $variant, $depth = 0 ) {
 634                 $startPos = 0;
 635                 $out = '';
 636                 $length = strlen( $text );
 637                 $shouldConvert = !$this->guessVariant( $text, $variant );
 638
 639                 while ( $startPos < $length ) {
 640                         $pos = strpos( $text, '-{', $startPos );
 641
 642                         if ( $pos === false ) {
 643                                 // No more markup, append final segment
 644                                 $fragment = substr( $text, $startPos );
 645                                 $out .= $shouldConvert ? $this->autoConvert( $fragment, $variant ) : $fragment;
 646                                 return $out;
 647                         }
 648
 649                         // Markup found
 650                         // Append initial segment
 651                         $fragment = substr( $text, $startPos, $pos - $startPos );
 652                         $out .= $shouldConvert ? $this->autoConvert( $fragment, $variant ) : $fragment;
 653
 654                         // Advance position
 655                         $startPos = $pos;
 656
 657                         // Do recursive conversion
 658                         $out .= $this->recursiveConvertRule( $text, $variant, $startPos, $depth + 1 );
 659                 }
 660
 661                 return $out;
 662         }
 663
 664         /**
 665          * Recursively convert text on the inside.
 666          *
 667          * @param string $text Text to be converted
 668          * @param string $variant The target variant code
 669          * @param int $startPos
 670          * @param int $depth Depth of recursion
 671          *
 672          * @throws MWException
 673          * @return string Converted text
 674          */
 675         protected function recursiveConvertRule( $text, $variant, &$startPos, $depth = 0 ) {
 676                 // Quick sanity check (no function calls)
 677                 if ( $text[$startPos] !== '-' || $text[$startPos + 1] !== '{' ) {
 678                         throw new MWException( __METHOD__ . ': invalid input string' );
 679                 }
 680
 681                 $startPos += 2;
 682                 $inner = '';
 683                 $warningDone = false;
 684                 $length = strlen( $text );
 685
 686                 while ( $startPos < $length ) {
 687                         $m = false;
 688                         preg_match( '/-\{|\}-/', $text, $m, PREG_OFFSET_CAPTURE, $startPos );
 689                         if ( !$m ) {
 690                                 // Unclosed rule
 691                                 break;
 692                         }
 693
 694                         $token = $m[0][0];
 695                         $pos = $m[0][1];
 696
 697                         // Markup found
 698                         // Append initial segment
 699                         $inner .= substr( $text, $startPos, $pos - $startPos );
 700
 701                         // Advance position
 702                         $startPos = $pos;
 703
 704                         switch ( $token ) {
 705                                 case '-{':
 706                                         // Check max depth
 707                                         if ( $depth >= $this->mMaxDepth ) {
 708                                                 $inner .= '-{';
 709                                                 if ( !$warningDone ) {
 710                                                         $inner .= '<span class="error">' .
 711                                                                 wfMessage( 'language-converter-depth-warning' )
 712                                                                         ->numParams( $this->mMaxDepth )->inContentLanguage()->text() .
 713                                                                 '</span>';
 714                                                         $warningDone = true;
 715                                                 }
 716                                                 $startPos += 2;
 717                                                 continue;
 718                                         }
 719                                         // Recursively parse another rule
 720                                         $inner .= $this->recursiveConvertRule( $text, $variant, $startPos, $depth + 1 );
 721                                         break;
 722                                 case '}-':
 723                                         // Apply the rule
 724                                         $startPos += 2;
 725                                         $rule = new ConverterRule( $inner, $this );
 726                                         $rule->parse( $variant );
 727                                         $this->applyManualConv( $rule );
 728                                         return $rule->getDisplay();
 729                                 default:
 730                                         throw new MWException( __METHOD__ . ': invalid regex match' );
 731                         }
 732                 }
 733
 734                 // Unclosed rule
 735                 if ( $startPos < $length ) {
 736                         $inner .= substr( $text, $startPos );
 737                 }
 738                 $startPos = $length;
 739                 return '-{' . $this->autoConvert( $inner, $variant );
 740         }
 741
 742         /**
 743          * If a language supports multiple variants, it is possible that
 744          * non-existing link in one variant actually exists in another variant.
 745          * This function tries to find it. See e.g. LanguageZh.php
 746          * The input parameters may be modified upon return
 747          *
 748          * @param string &$link The name of the link
 749          * @param Title &$nt The title object of the link
 750          * @param bool $ignoreOtherCond To disable other conditions when
 751          *   we need to transclude a template or update a category's link
 752          */
 753         public function findVariantLink( &$link, &$nt, $ignoreOtherCond = false ) {
 754                 # If the article has already existed, there is no need to
 755                 # check it again, otherwise it may cause a fault.
 756                 if ( is_object( $nt ) && $nt->exists() ) {
 757                         return;
 758                 }
 759
 760                 global $wgDisableLangConversion, $wgDisableTitleConversion, $wgRequest;
 761                 $isredir = $wgRequest->getText( 'redirect', 'yes' );
 762                 $action = $wgRequest->getText( 'action' );
 763                 if ( $action == 'edit' && $wgRequest->getBool( 'redlink' ) ) {
 764                         $action = 'view';
 765                 }
 766                 $linkconvert = $wgRequest->getText( 'linkconvert', 'yes' );
 767                 $disableLinkConversion = $wgDisableLangConversion
 768                         || $wgDisableTitleConversion;
 769                 $linkBatch = new LinkBatch();
 770
 771                 $ns = NS_MAIN;
 772
 773                 if ( $disableLinkConversion ||
 774                         ( !$ignoreOtherCond &&
 775                                 ( $isredir == 'no'
 776                                         || $action == 'edit'
 777                                         || $action == 'submit'
 778                                         || $linkconvert == 'no' ) ) ) {
 779                         return;
 780                 }
 781
 782                 if ( is_object( $nt ) ) {
 783                         $ns = $nt->getNamespace();
 784                 }
 785
 786                 $variants = $this->autoConvertToAllVariants( $link );
 787                 if ( !$variants ) { // give up
 788                         return;
 789                 }
 790
 791                 $titles = [];
 792
 793                 foreach ( $variants as $v ) {
 794                         if ( $v != $link ) {
 795                                 $varnt = Title::newFromText( $v, $ns );
 796                                 if ( !is_null( $varnt ) ) {
 797                                         $linkBatch->addObj( $varnt );
 798                                         $titles[] = $varnt;
 799                                 }
 800                         }
 801                 }
 802
 803                 // fetch all variants in single query
 804                 $linkBatch->execute();
 805
 806                 foreach ( $titles as $varnt ) {
 807                         if ( $varnt->getArticleID() > 0 ) {
 808                                 $nt = $varnt;
 809                                 $link = $varnt->getText();
 810                                 break;
 811                         }
 812                 }
 813         }
 814
 815         /**
 816          * Returns language specific hash options.
 817          *
 818          * @return string
 819          */
 820         public function getExtraHashOptions() {
 821                 $variant = $this->getPreferredVariant();
 822
 823                 return '!' . $variant;
 824         }
 825
 826         /**
 827          * Guess if a text is written in a variant. This should be implemented in subclasses.
 828          *
 829          * @param string $text The text to be checked
 830          * @param string $variant Language code of the variant to be checked for
 831          * @return bool True if $text appears to be written in $variant, false if not
 832          *
 833          * @author Nikola Smolenski <smolensk@eunet.rs>
 834          * @since 1.19
 835          */
 836         public function guessVariant( $text, $variant ) {
 837                 return false;
 838         }
 839
 840         /**
 841          * Load default conversion tables.
 842          * This method must be implemented in derived class.
 843          *
 844          * @private
 845          * @throws MWException
 846          */
 847         function loadDefaultTables() {
 848                 $class = static::class;
 849                 throw new MWException( "Must implement loadDefaultTables() method in class $class" );
 850         }
 851
 852         /**
 853          * Load conversion tables either from the cache or the disk.
 854          * @private
 855          * @param bool $fromCache Load from memcached? Defaults to true.
 856          */
 857         function loadTables( $fromCache = true ) {
 858                 global $wgLanguageConverterCacheType;
 859
 860                 if ( $this->mTablesLoaded ) {
 861                         return;
 862                 }
 863
 864                 $this->mTablesLoaded = true;
 865                 $this->mTables = false;
 866                 $cache = ObjectCache::getInstance( $wgLanguageConverterCacheType );
 867                 if ( $fromCache ) {
 868                         wfProfileIn( __METHOD__ . '-cache' );
 869                         $this->mTables = $cache->get( $this->mCacheKey );
 870                         wfProfileOut( __METHOD__ . '-cache' );
 871                 }
 872                 if ( !$this->mTables || !array_key_exists( self::CACHE_VERSION_KEY, $this->mTables ) ) {
 873                         wfProfileIn( __METHOD__ . '-recache' );
 874                         // not in cache, or we need a fresh reload.
 875                         // We will first load the default tables
 876                         // then update them using things in MediaWiki:Conversiontable/*
 877                         $this->loadDefaultTables();
 878                         foreach ( $this->mVariants as $var ) {
 879                                 $cached = $this->parseCachedTable( $var );
 880                                 $this->mTables[$var]->mergeArray( $cached );
 881                         }
 882
 883                         $this->postLoadTables();
 884                         $this->mTables[self::CACHE_VERSION_KEY] = true;
 885
 886                         $cache->set( $this->mCacheKey, $this->mTables, 43200 );
 887                         wfProfileOut( __METHOD__ . '-recache' );
 888                 }
 889         }
 890
 891         /**
 892          * Hook for post processing after conversion tables are loaded.
 893          */
 894         function postLoadTables() {
 895         }
 896
 897         /**
 898          * Reload the conversion tables.
 899          *
 900          * @private
 901          */
 902         function reloadTables() {
 903                 if ( $this->mTables ) {
 904                         unset( $this->mTables );
 905                 }
 906
 907                 $this->mTablesLoaded = false;
 908                 $this->loadTables( false );
 909         }
 910
 911         /**
 912          * Parse the conversion table stored in the cache.
 913          *
 914          * The tables should be in blocks of the following form:
 915          *              -{
 916          *                      word => word ;
 917          *                      word => word ;
 918          *                      ...
 919          *              }-
 920          *
 921          * To make the tables more manageable, subpages are allowed
 922          * and will be parsed recursively if $recursive == true.
 923          *
 924          * @param string $code Language code
 925          * @param string $subpage Subpage name
 926          * @param bool $recursive Parse subpages recursively? Defaults to true.
 927          *
 928          * @return array
 929          */
 930         function parseCachedTable( $code, $subpage = '', $recursive = true ) {
 931                 static $parsed = [];
 932
 933                 $key = 'Conversiontable/' . $code;
 934                 if ( $subpage ) {
 935                         $key .= '/' . $subpage;
 936                 }
 937                 if ( array_key_exists( $key, $parsed ) ) {
 938                         return [];
 939                 }
 940
 941                 $parsed[$key] = true;
 942
 943                 if ( $subpage === '' ) {
 944                         $txt = MessageCache::singleton()->getMsgFromNamespace( $key, $code );
 945                 } else {
 946                         $txt = false;
 947                         $title = Title::makeTitleSafe( NS_MEDIAWIKI, $key );
 948                         if ( $title && $title->exists() ) {
 949                                 $revision = Revision::newFromTitle( $title );
 950                                 if ( $revision ) {
 951                                         if ( $revision->getContentModel() == CONTENT_MODEL_WIKITEXT ) {
 952                                                 $txt = $revision->getContent( Revision::RAW )->getNativeData();
 953                                         }
 954
 955                                         // @todo in the future, use a specialized content model, perhaps based on json!
 956                                 }
 957                         }
 958                 }
 959
 960                 # Nothing to parse if there's no text
 961                 if ( $txt === false || $txt === null || $txt === '' ) {
 962                         return [];
 963                 }
 964
 965                 // get all subpage links of the form
 966                 // [[MediaWiki:Conversiontable/zh-xx/...|...]]
 967                 $linkhead = $this->mLangObj->getNsText( NS_MEDIAWIKI ) .
 968                         ':Conversiontable';
 969                 $subs = StringUtils::explode( '[[', $txt );
 970                 $sublinks = [];
 971                 foreach ( $subs as $sub ) {
 972                         $link = explode( ']]', $sub, 2 );
 973                         if ( count( $link ) != 2 ) {
 974                                 continue;
 975                         }
 976                         $b = explode( '|', $link[0], 2 );
 977                         $b = explode( '/', trim( $b[0] ), 3 );
 978                         if ( count( $b ) == 3 ) {
 979                                 $sublink = $b[2];
 980                         } else {
 981                                 $sublink = '';
 982                         }
 983
 984                         if ( $b[0] == $linkhead && $b[1] == $code ) {
 985                                 $sublinks[] = $sublink;
 986                         }
 987                 }
 988
 989                 // parse the mappings in this page
 990                 $blocks = StringUtils::explode( '-{', $txt );
 991                 $ret = [];
 992                 $first = true;
 993                 foreach ( $blocks as $block ) {
 994                         if ( $first ) {
 995                                 // Skip the part before the first -{
 996                                 $first = false;
 997                                 continue;
 998                         }
 999                         $mappings = explode( '}-', $block, 2 )[0];
1000                         $stripped = str_replace( [ "'", '"', '*', '#' ], '', $mappings );
1001                         $table = StringUtils::explode( ';', $stripped );
1002                         foreach ( $table as $t ) {
1003                                 $m = explode( '=>', $t, 3 );
1004                                 if ( count( $m ) != 2 ) {
1005                                         continue;
1006                                 }
1007                                 // trim any trailling comments starting with '//'
1008                                 $tt = explode( '//', $m[1], 2 );
1009                                 $ret[trim( $m[0] )] = trim( $tt[0] );
1010                         }
1011                 }
1012
1013                 // recursively parse the subpages
1014                 if ( $recursive ) {
1015                         foreach ( $sublinks as $link ) {
1016                                 $s = $this->parseCachedTable( $code, $link, $recursive );
1017                                 $ret = $s + $ret;
1018                         }
1019                 }
1020
1021                 if ( $this->mUcfirst ) {
1022                         foreach ( $ret as $k => $v ) {
1023                                 $ret[$this->mLangObj->ucfirst( $k )] = $this->mLangObj->ucfirst( $v );
1024                         }
1025                 }
1026                 return $ret;
1027         }
1028
1029         /**
1030          * Enclose a string with the "no conversion" tag. This is used by
1031          * various functions in the Parser.
1032          *
1033          * @param string $text Text to be tagged for no conversion
1034          * @param bool $noParse Unused
1035          * @return string The tagged text
1036          */
1037         public function markNoConversion( $text, $noParse = false ) {
1038                 # don't mark if already marked
1039                 if ( strpos( $text, '-{' ) || strpos( $text, '}-' ) ) {
1040                         return $text;
1041                 }
1042
1043                 $ret = "-{R|$text}-";
1044                 return $ret;
1045         }
1046
1047         /**
1048          * Convert the sorting key for category links. This should make different
1049          * keys that are variants of each other map to the same key.
1050          *
1051          * @param string $key
1052          *
1053          * @return string
1054          */
1055         function convertCategoryKey( $key ) {
1056                 return $key;
1057         }
1058
1059         /**
1060          * Refresh the cache of conversion tables when
1061          * MediaWiki:Conversiontable* is updated.
1062          *
1063          * @param Title $titleobj The Title of the page being updated
1064          */
1065         public function updateConversionTable( Title $titleobj ) {
1066                 if ( $titleobj->getNamespace() == NS_MEDIAWIKI ) {
1067                         $title = $titleobj->getDBkey();
1068                         $t = explode( '/', $title, 3 );
1069                         $c = count( $t );
1070                         if ( $c > 1 && $t[0] == 'Conversiontable' ) {
1071                                 if ( $this->validateVariant( $t[1] ) ) {
1072                                         $this->reloadTables();
1073                                 }
1074                         }
1075                 }
1076         }
1077
1078         /**
1079          * Get the cached separator pattern for ConverterRule::parseRules()
1080          * @return string
1081          */
1082         function getVarSeparatorPattern() {
1083                 if ( is_null( $this->mVarSeparatorPattern ) ) {
1084                         // varsep_pattern for preg_split:
1085                         // text should be splited by ";" only if a valid variant
1086                         // name exist after the markup, for example:
1087                         //  -{zh-hans:<span style="font-size:120%;">xxx</span>;zh-hant:\
1088                         //  <span style="font-size:120%;">yyy</span>;}-
1089                         // we should split it as:
1090                         //  [
1091                         //    [0] => 'zh-hans:<span style="font-size:120%;">xxx</span>'
1092                         //    [1] => 'zh-hant:<span style="font-size:120%;">yyy</span>'
1093                         //    [2] => ''
1094                         //  ]
1095                         $pat = '/;\s*(?=';
1096                         foreach ( $this->mVariants as $variant ) {
1097                                 // zh-hans:xxx;zh-hant:yyy
1098                                 $pat .= $variant . '\s*:|';
1099                                 // xxx=>zh-hans:yyy; xxx=>zh-hant:zzz
1100                                 $pat .= '[^;]*?=>\s*' . $variant . '\s*:|';
1101                         }
1102                         $pat .= '\s*$)/';
1103                         $this->mVarSeparatorPattern = $pat;
1104                 }
1105                 return $this->mVarSeparatorPattern;
1106         }
1107 }