languages/LanguageConverter.php

   1 <?php
   2 /**
   3  * This program is free software; you can redistribute it and/or modify
   4  * it under the terms of the GNU General Public License as published by
   5  * the Free Software Foundation; either version 2 of the License, or
   6  * (at your option) any later version.
   7  *
   8  * This program is distributed in the hope that it will be useful,
   9  * but WITHOUT ANY WARRANTY; without even the implied warranty of
  10  * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the
  11  * GNU General Public License for more details.
  12  *
  13  * You should have received a copy of the GNU General Public License along
  14  * with this program; if not, write to the Free Software Foundation, Inc.,
  15  * 51 Franklin Street, Fifth Floor, Boston, MA 02110-1301, USA.
  16  * http://www.gnu.org/copyleft/gpl.html
  17  *
  18  * @file
  19  * @ingroup Language
  20  */
  21 use MediaWiki\MediaWikiServices;
  22
  23 use MediaWiki\Logger\LoggerFactory;
  24
  25 /**
  26  * Base class for language conversion.
  27  * @ingroup Language
  28  *
  29  * @author Zhengzhu Feng <zhengzhu@gmail.com>
  30  * @author fdcn <fdcn64@gmail.com>
  31  * @author shinjiman <shinjiman@gmail.com>
  32  * @author PhiLiP <philip.npc@gmail.com>
  33  */
  34 class LanguageConverter {
  35         /**
  36          * languages supporting variants
  37          * @since 1.20
  38          * @var array
  39          */
  40         static public $languagesWithVariants = [
  41                 'en',
  42                 'crh',
  43                 'gan',
  44                 'iu',
  45                 'kk',
  46                 'ku',
  47                 'shi',
  48                 'sr',
  49                 'tg',
  50                 'uz',
  51                 'zh',
  52         ];
  53
  54         public $mMainLanguageCode;
  55
  56         /**
  57          * @var string[]
  58          */
  59         public $mVariants;
  60         public $mVariantFallbacks;
  61         public $mVariantNames;
  62         public $mTablesLoaded = false;
  63         public $mTables;
  64         // 'bidirectional' 'unidirectional' 'disable' for each variant
  65         public $mManualLevel;
  66
  67         public $mLangObj;
  68         public $mFlags;
  69         public $mDescCodeSep = ':', $mDescVarSep = ';';
  70         public $mUcfirst = false;
  71         public $mConvRuleTitle = false;
  72         public $mURLVariant;
  73         public $mUserVariant;
  74         public $mHeaderVariant;
  75         public $mMaxDepth = 10;
  76         public $mVarSeparatorPattern;
  77
  78         const CACHE_VERSION_KEY = 'VERSION 7';
  79
  80         /**
  81          * @param Language $langobj
  82          * @param string $maincode The main language code of this language
  83          * @param string[] $variants The supported variants of this language
  84          * @param array $variantfallbacks The fallback language of each variant
  85          * @param array $flags Defining the custom strings that maps to the flags
  86          * @param array $manualLevel Limit for supported variants
  87          */
  88         public function __construct( $langobj, $maincode, $variants = [],
  89                                                                 $variantfallbacks = [], $flags = [],
  90                                                                 $manualLevel = [] ) {
  91                 global $wgDisabledVariants;
  92                 $this->mLangObj = $langobj;
  93                 $this->mMainLanguageCode = $maincode;
  94                 $this->mVariants = array_diff( $variants, $wgDisabledVariants );
  95                 $this->mVariantFallbacks = $variantfallbacks;
  96                 $this->mVariantNames = Language::fetchLanguageNames();
  97                 $defaultflags = [
  98                         // 'S' show converted text
  99                         // '+' add rules for alltext
 100                         // 'E' the gave flags is error
 101                         // these flags above are reserved for program
 102                         'A' => 'A',   // add rule for convert code (all text convert)
 103                         'T' => 'T',   // title convert
 104                         'R' => 'R',   // raw content
 105                         'D' => 'D',   // convert description (subclass implement)
 106                         '-' => '-',   // remove convert (not implement)
 107                         'H' => 'H',   // add rule for convert code (but no display in placed code)
 108                         'N' => 'N',   // current variant name
 109                 ];
 110                 $this->mFlags = array_merge( $defaultflags, $flags );
 111                 foreach ( $this->mVariants as $v ) {
 112                         if ( array_key_exists( $v, $manualLevel ) ) {
 113                                 $this->mManualLevel[$v] = $manualLevel[$v];
 114                         } else {
 115                                 $this->mManualLevel[$v] = 'bidirectional';
 116                         }
 117                         $this->mFlags[$v] = $v;
 118                 }
 119         }
 120
 121         /**
 122          * Get all valid variants.
 123          * Call this instead of using $this->mVariants directly.
 124          *
 125          * @return string[] Contains all valid variants
 126          */
 127         public function getVariants() {
 128                 return $this->mVariants;
 129         }
 130
 131         /**
 132          * In case some variant is not defined in the markup, we need
 133          * to have some fallback. For example, in zh, normally people
 134          * will define zh-hans and zh-hant, but less so for zh-sg or zh-hk.
 135          * when zh-sg is preferred but not defined, we will pick zh-hans
 136          * in this case. Right now this is only used by zh.
 137          *
 138          * @param string $variant The language code of the variant
 139          * @return string|array The code of the fallback language or the
 140          *   main code if there is no fallback
 141          */
 142         public function getVariantFallbacks( $variant ) {
 143                 if ( isset( $this->mVariantFallbacks[$variant] ) ) {
 144                         return $this->mVariantFallbacks[$variant];
 145                 }
 146                 return $this->mMainLanguageCode;
 147         }
 148
 149         /**
 150          * Get the title produced by the conversion rule.
 151          * @return string The converted title text
 152          */
 153         public function getConvRuleTitle() {
 154                 return $this->mConvRuleTitle;
 155         }
 156
 157         /**
 158          * Get preferred language variant.
 159          * @return string The preferred language code
 160          */
 161         public function getPreferredVariant() {
 162                 global $wgDefaultLanguageVariant, $wgUser;
 163
 164                 $req = $this->getURLVariant();
 165
 166                 Hooks::run( 'GetLangPreferredVariant', [ &$req ] );
 167
 168                 if ( $wgUser->isSafeToLoad() && $wgUser->isLoggedIn() && !$req ) {
 169                         $req = $this->getUserVariant();
 170                 } elseif ( !$req ) {
 171                         $req = $this->getHeaderVariant();
 172                 }
 173
 174                 if ( $wgDefaultLanguageVariant && !$req ) {
 175                         $req = $this->validateVariant( $wgDefaultLanguageVariant );
 176                 }
 177
 178                 $req = $this->validateVariant( $req );
 179
 180                 // This function, unlike the other get*Variant functions, is
 181                 // not memoized (i.e. there return value is not cached) since
 182                 // new information might appear during processing after this
 183                 // is first called.
 184                 if ( $req ) {
 185                         return $req;
 186                 }
 187                 return $this->mMainLanguageCode;
 188         }
 189
 190         /**
 191          * Get default variant.
 192          * This function would not be affected by user's settings
 193          * @return string The default variant code
 194          */
 195         public function getDefaultVariant() {
 196                 global $wgDefaultLanguageVariant;
 197
 198                 $req = $this->getURLVariant();
 199
 200                 if ( !$req ) {
 201                         $req = $this->getHeaderVariant();
 202                 }
 203
 204                 if ( $wgDefaultLanguageVariant && !$req ) {
 205                         $req = $this->validateVariant( $wgDefaultLanguageVariant );
 206                 }
 207
 208                 if ( $req ) {
 209                         return $req;
 210                 }
 211                 return $this->mMainLanguageCode;
 212         }
 213
 214         /**
 215          * Validate the variant
 216          * @param string|null $variant The variant to validate
 217          * @return mixed Returns the variant if it is valid, null otherwise
 218          */
 219         public function validateVariant( $variant = null ) {
 220                 if ( $variant === null ) {
 221                         return null;
 222                 }
 223                 // Our internal variants are always lower-case; the variant we
 224                 // are validating may have mixed case.
 225                 $variant = LanguageCode::replaceDeprecatedCodes( strtolower( $variant ) );
 226                 if ( in_array( $variant, $this->mVariants ) ) {
 227                         return $variant;
 228                 }
 229                 // Browsers are supposed to use BCP 47 standard in the
 230                 // Accept-Language header, but not all of our internal
 231                 // mediawiki variant codes are BCP 47.  Map BCP 47 code
 232                 // to our internal code.
 233                 foreach ( $this->mVariants as $v ) {
 234                         // Case-insensitive match (BCP 47 is mixed case)
 235                         if ( strtolower( LanguageCode::bcp47( $v ) ) === $variant ) {
 236                                 return $v;
 237                         }
 238                 }
 239                 return null;
 240         }
 241
 242         /**
 243          * Get the variant specified in the URL
 244          *
 245          * @return mixed Variant if one found, null otherwise
 246          */
 247         public function getURLVariant() {
 248                 global $wgRequest;
 249
 250                 if ( $this->mURLVariant ) {
 251                         return $this->mURLVariant;
 252                 }
 253
 254                 // see if the preference is set in the request
 255                 $ret = $wgRequest->getText( 'variant' );
 256
 257                 if ( !$ret ) {
 258                         $ret = $wgRequest->getVal( 'uselang' );
 259                 }
 260
 261                 $this->mURLVariant = $this->validateVariant( $ret );
 262                 return $this->mURLVariant;
 263         }
 264
 265         /**
 266          * Determine if the user has a variant set.
 267          *
 268          * @return mixed Variant if one found, null otherwise
 269          */
 270         protected function getUserVariant() {
 271                 global $wgUser, $wgContLang;
 272
 273                 // memoizing this function wreaks havoc on parserTest.php
 274                 /*
 275                 if ( $this->mUserVariant ) {
 276                         return $this->mUserVariant;
 277                 }
 278                 */
 279
 280                 // Get language variant preference from logged in users
 281                 // Don't call this on stub objects because that causes infinite
 282                 // recursion during initialisation
 283                 if ( !$wgUser->isSafeToLoad() ) {
 284                         return false;
 285                 }
 286                 if ( $wgUser->isLoggedIn() ) {
 287                         if ( $this->mMainLanguageCode == $wgContLang->getCode() ) {
 288                                 $ret = $wgUser->getOption( 'variant' );
 289                         } else {
 290                                 $ret = $wgUser->getOption( 'variant-' . $this->mMainLanguageCode );
 291                         }
 292                 } else {
 293                         // figure out user lang without constructing wgLang to avoid
 294                         // infinite recursion
 295                         $ret = $wgUser->getOption( 'language' );
 296                 }
 297
 298                 $this->mUserVariant = $this->validateVariant( $ret );
 299                 return $this->mUserVariant;
 300         }
 301
 302         /**
 303          * Determine the language variant from the Accept-Language header.
 304          *
 305          * @return mixed Variant if one found, null otherwise
 306          */
 307         protected function getHeaderVariant() {
 308                 global $wgRequest;
 309
 310                 if ( $this->mHeaderVariant ) {
 311                         return $this->mHeaderVariant;
 312                 }
 313
 314                 // See if some supported language variant is set in the
 315                 // HTTP header.
 316                 $languages = array_keys( $wgRequest->getAcceptLang() );
 317                 if ( empty( $languages ) ) {
 318                         return null;
 319                 }
 320
 321                 $fallbackLanguages = [];
 322                 foreach ( $languages as $language ) {
 323                         $this->mHeaderVariant = $this->validateVariant( $language );
 324                         if ( $this->mHeaderVariant ) {
 325                                 break;
 326                         }
 327
 328                         // To see if there are fallbacks of current language.
 329                         // We record these fallback variants, and process
 330                         // them later.
 331                         $fallbacks = $this->getVariantFallbacks( $language );
 332                         if ( is_string( $fallbacks ) && $fallbacks !== $this->mMainLanguageCode ) {
 333                                 $fallbackLanguages[] = $fallbacks;
 334                         } elseif ( is_array( $fallbacks ) ) {
 335                                 $fallbackLanguages =
 336                                         array_merge( $fallbackLanguages, $fallbacks );
 337                         }
 338                 }
 339
 340                 if ( !$this->mHeaderVariant ) {
 341                         // process fallback languages now
 342                         $fallback_languages = array_unique( $fallbackLanguages );
 343                         foreach ( $fallback_languages as $language ) {
 344                                 $this->mHeaderVariant = $this->validateVariant( $language );
 345                                 if ( $this->mHeaderVariant ) {
 346                                         break;
 347                                 }
 348                         }
 349                 }
 350
 351                 return $this->mHeaderVariant;
 352         }
 353
 354         /**
 355          * Dictionary-based conversion.
 356          * This function would not parse the conversion rules.
 357          * If you want to parse rules, try to use convert() or
 358          * convertTo().
 359          *
 360          * @param string $text The text to be converted
 361          * @param bool|string $toVariant The target language code
 362          * @return string The converted text
 363          */
 364         public function autoConvert( $text, $toVariant = false ) {
 365                 $this->loadTables();
 366
 367                 if ( !$toVariant ) {
 368                         $toVariant = $this->getPreferredVariant();
 369                         if ( !$toVariant ) {
 370                                 return $text;
 371                         }
 372                 }
 373
 374                 if ( $this->guessVariant( $text, $toVariant ) ) {
 375                         return $text;
 376                 }
 377                 /* we convert everything except:
 378                    1. HTML markups (anything between < and >)
 379                    2. HTML entities
 380                    3. placeholders created by the parser
 381                    IMPORTANT: Beware of failure from pcre.backtrack_limit (T124404).
 382                    Minimize use of backtracking where possible.
 383                 */
 384                 $marker = '|' . Parser::MARKER_PREFIX . '[^\x7f]++\x7f';
 385
 386                 // this one is needed when the text is inside an HTML markup
 387                 $htmlfix = '|<[^>\004]++(?=\004$)|^[^<>]*+>';
 388
 389                 // Optimize for the common case where these tags have
 390                 // few or no children. Thus try and possesively get as much as
 391                 // possible, and only engage in backtracking when we hit a '<'.
 392
 393                 // disable convert to variants between <code> tags
 394                 $codefix = '<code>[^<]*+(?:(?:(?!<\/code>).)[^<]*+)*+<\/code>|';
 395                 // disable conversion of <script> tags
 396                 $scriptfix = '<script[^>]*+>[^<]*+(?:(?:(?!<\/script>).)[^<]*+)*+<\/script>|';
 397                 // disable conversion of <pre> tags
 398                 $prefix = '<pre[^>]*+>[^<]*+(?:(?:(?!<\/pre>).)[^<]*+)*+<\/pre>|';
 399                 // The "|.*+)" at the end, is in case we missed some part of html syntax,
 400                 // we will fail securely (hopefully) by matching the rest of the string.
 401                 $htmlFullTag = '<(?:[^>=]*+(?>[^>=]*+=\s*+(?:"[^"]*"|\'[^\']*\'|[^\'">\s]*+))*+[^>=]*+>|.*+)|';
 402
 403                 $reg = '/' . $codefix . $scriptfix . $prefix . $htmlFullTag .
 404                         '&[a-zA-Z#][a-z0-9]++;' . $marker . $htmlfix . '|\004$/s';
 405                 $startPos = 0;
 406                 $sourceBlob = '';
 407                 $literalBlob = '';
 408
 409                 // Guard against delimiter nulls in the input
 410                 // (should never happen: see T159174)
 411                 $text = str_replace( "\000", '', $text );
 412                 $text = str_replace( "\004", '', $text );
 413
 414                 $markupMatches = null;
 415                 $elementMatches = null;
 416
 417                 // We add a marker (\004) at the end of text, to ensure we always match the
 418                 // entire text (Otherwise, pcre.backtrack_limit might cause silent failure)
 419                 while ( $startPos < strlen( $text ) ) {
 420                         if ( preg_match( $reg, $text . "\004", $markupMatches, PREG_OFFSET_CAPTURE, $startPos ) ) {
 421                                 $elementPos = $markupMatches[0][1];
 422                                 $element = $markupMatches[0][0];
 423                                 if ( $element === "\004" ) {
 424                                         // We hit the end.
 425                                         $elementPos = strlen( $text );
 426                                         $element = '';
 427                                 } elseif ( substr( $element, -1 ) === "\004" ) {
 428                                         // This can sometimes happen if we have
 429                                         // unclosed html tags (For example
 430                                         // when converting a title attribute
 431                                         // during a recursive call that contains
 432                                         // a &lt; e.g. <div title="&lt;">.
 433                                         $element = substr( $element, 0, -1 );
 434                                 }
 435                         } else {
 436                                 // If we hit here, then Language Converter could be tricked
 437                                 // into doing an XSS, so we refuse to translate.
 438                                 // If non-crazy input manages to reach this code path,
 439                                 // we should consider it a bug.
 440                                 $log = LoggerFactory::getInstance( 'languageconverter' );
 441                                 $log->error( "Hit pcre.backtrack_limit in " . __METHOD__
 442                                         . ". Disabling language conversion for this page.",
 443                                         [
 444                                                 "method" => __METHOD__,
 445                                                 "variant" => $toVariant,
 446                                                 "startOfText" => substr( $text, 0, 500 )
 447                                         ]
 448                                 );
 449                                 return $text;
 450                         }
 451                         // Queue the part before the markup for translation in a batch
 452                         $sourceBlob .= substr( $text, $startPos, $elementPos - $startPos ) . "\000";
 453
 454                         // Advance to the next position
 455                         $startPos = $elementPos + strlen( $element );
 456
 457                         // Translate any alt or title attributes inside the matched element
 458                         if ( $element !== ''
 459                                 && preg_match( '/^(<[^>\s]*+)\s([^>]*+)(.*+)$/', $element, $elementMatches )
 460                         ) {
 461                                 // FIXME, this decodes entities, so if you have something
 462                                 // like <div title="foo&lt;bar"> the bar won't get
 463                                 // translated since after entity decoding it looks like
 464                                 // unclosed html and we call this method recursively
 465                                 // on attributes.
 466                                 $attrs = Sanitizer::decodeTagAttributes( $elementMatches[2] );
 467                                 // Ensure self-closing tags stay self-closing.
 468                                 $close = substr( $elementMatches[2], -1 ) === '/' ? ' /' : '';
 469                                 $changed = false;
 470                                 foreach ( [ 'title', 'alt' ] as $attrName ) {
 471                                         if ( !isset( $attrs[$attrName] ) ) {
 472                                                 continue;
 473                                         }
 474                                         $attr = $attrs[$attrName];
 475                                         // Don't convert URLs
 476                                         if ( !strpos( $attr, '://' ) ) {
 477                                                 $attr = $this->recursiveConvertTopLevel( $attr, $toVariant );
 478                                         }
 479
 480                                         if ( $attr !== $attrs[$attrName] ) {
 481                                                 $attrs[$attrName] = $attr;
 482                                                 $changed = true;
 483                                         }
 484                                 }
 485                                 if ( $changed ) {
 486                                         $element = $elementMatches[1] . Html::expandAttributes( $attrs ) .
 487                                                 $close . $elementMatches[3];
 488                                 }
 489                         }
 490                         $literalBlob .= $element . "\000";
 491                 }
 492
 493                 // Do the main translation batch
 494                 $translatedBlob = $this->translate( $sourceBlob, $toVariant );
 495
 496                 // Put the output back together
 497                 $translatedIter = StringUtils::explode( "\000", $translatedBlob );
 498                 $literalIter = StringUtils::explode( "\000", $literalBlob );
 499                 $output = '';
 500                 while ( $translatedIter->valid() && $literalIter->valid() ) {
 501                         $output .= $translatedIter->current();
 502                         $output .= $literalIter->current();
 503                         $translatedIter->next();
 504                         $literalIter->next();
 505                 }
 506
 507                 return $output;
 508         }
 509
 510         /**
 511          * Translate a string to a variant.
 512          * Doesn't parse rules or do any of that other stuff, for that use
 513          * convert() or convertTo().
 514          *
 515          * @param string $text Text to convert
 516          * @param string $variant Variant language code
 517          * @return string Translated text
 518          */
 519         public function translate( $text, $variant ) {
 520                 // If $text is empty or only includes spaces, do nothing
 521                 // Otherwise translate it
 522                 if ( trim( $text ) ) {
 523                         $this->loadTables();
 524                         $text = $this->mTables[$variant]->replace( $text );
 525                 }
 526                 return $text;
 527         }
 528
 529         /**
 530          * Call translate() to convert text to all valid variants.
 531          *
 532          * @param string $text The text to be converted
 533          * @return array Variant => converted text
 534          */
 535         public function autoConvertToAllVariants( $text ) {
 536                 $this->loadTables();
 537
 538                 $ret = [];
 539                 foreach ( $this->mVariants as $variant ) {
 540                         $ret[$variant] = $this->translate( $text, $variant );
 541                 }
 542
 543                 return $ret;
 544         }
 545
 546         /**
 547          * Apply manual conversion rules.
 548          *
 549          * @param ConverterRule $convRule
 550          */
 551         protected function applyManualConv( $convRule ) {
 552                 // Use syntax -{T|zh-cn:TitleCN; zh-tw:TitleTw}- to custom
 553                 // title conversion.
 554                 // T26072: $mConvRuleTitle was overwritten by other manual
 555                 // rule(s) not for title, this breaks the title conversion.
 556                 $newConvRuleTitle = $convRule->getTitle();
 557                 if ( $newConvRuleTitle ) {
 558                         // So I add an empty check for getTitle()
 559                         $this->mConvRuleTitle = $newConvRuleTitle;
 560                 }
 561
 562                 // merge/remove manual conversion rules to/from global table
 563                 $convTable = $convRule->getConvTable();
 564                 $action = $convRule->getRulesAction();
 565                 foreach ( $convTable as $variant => $pair ) {
 566                         $v = $this->validateVariant( $variant );
 567                         if ( !$v ) {
 568                                 continue;
 569                         }
 570
 571                         if ( $action == 'add' ) {
 572                                 // More efficient than array_merge(), about 2.5 times.
 573                                 foreach ( $pair as $from => $to ) {
 574                                         $this->mTables[$v]->setPair( $from, $to );
 575                                 }
 576                         } elseif ( $action == 'remove' ) {
 577                                 $this->mTables[$v]->removeArray( $pair );
 578                         }
 579                 }
 580         }
 581
 582         /**
 583          * Auto convert a Title object to a readable string in the
 584          * preferred variant.
 585          *
 586          * @param Title $title A object of Title
 587          * @return string Converted title text
 588          */
 589         public function convertTitle( $title ) {
 590                 $variant = $this->getPreferredVariant();
 591                 $index = $title->getNamespace();
 592                 if ( $index !== NS_MAIN ) {
 593                         $text = $this->convertNamespace( $index, $variant ) . ':';
 594                 } else {
 595                         $text = '';
 596                 }
 597                 $text .= $this->translate( $title->getText(), $variant );
 598                 return $text;
 599         }
 600
 601         /**
 602          * Get the namespace display name in the preferred variant.
 603          *
 604          * @param int $index Namespace id
 605          * @param string|null $variant Variant code or null for preferred variant
 606          * @return string Namespace name for display
 607          */
 608         public function convertNamespace( $index, $variant = null ) {
 609                 if ( $index === NS_MAIN ) {
 610                         return '';
 611                 }
 612
 613                 if ( $variant === null ) {
 614                         $variant = $this->getPreferredVariant();
 615                 }
 616
 617                 $cache = MediaWikiServices::getInstance()->getLocalServerObjectCache();
 618                 $key = $cache->makeKey( 'languageconverter', 'namespace-text', $index, $variant );
 619                 $nsVariantText = $cache->get( $key );
 620                 if ( $nsVariantText !== false ) {
 621                         return $nsVariantText;
 622                 }
 623
 624                 // First check if a message gives a converted name in the target variant.
 625                 $nsConvMsg = wfMessage( 'conversion-ns' . $index )->inLanguage( $variant );
 626                 if ( $nsConvMsg->exists() ) {
 627                         $nsVariantText = $nsConvMsg->plain();
 628                 }
 629
 630                 // Then check if a message gives a converted name in content language
 631                 // which needs extra translation to the target variant.
 632                 if ( $nsVariantText === false ) {
 633                         $nsConvMsg = wfMessage( 'conversion-ns' . $index )->inContentLanguage();
 634                         if ( $nsConvMsg->exists() ) {
 635                                 $nsVariantText = $this->translate( $nsConvMsg->plain(), $variant );
 636                         }
 637                 }
 638
 639                 if ( $nsVariantText === false ) {
 640                         // No message exists, retrieve it from the target variant's namespace names.
 641                         $langObj = $this->mLangObj->factory( $variant );
 642                         $nsVariantText = $langObj->getFormattedNsText( $index );
 643                 }
 644
 645                 $cache->set( $key, $nsVariantText, 60 );
 646
 647                 return $nsVariantText;
 648         }
 649
 650         /**
 651          * Convert text to different variants of a language. The automatic
 652          * conversion is done in autoConvert(). Here we parse the text
 653          * marked with -{}-, which specifies special conversions of the
 654          * text that can not be accomplished in autoConvert().
 655          *
 656          * Syntax of the markup:
 657          * -{code1:text1;code2:text2;...}-  or
 658          * -{flags|code1:text1;code2:text2;...}-  or
 659          * -{text}- in which case no conversion should take place for text
 660          *
 661          * @param string $text Text to be converted
 662          * @return string Converted text
 663          */
 664         public function convert( $text ) {
 665                 $variant = $this->getPreferredVariant();
 666                 return $this->convertTo( $text, $variant );
 667         }
 668
 669         /**
 670          * Same as convert() except a extra parameter to custom variant.
 671          *
 672          * @param string $text Text to be converted
 673          * @param string $variant The target variant code
 674          * @return string Converted text
 675          */
 676         public function convertTo( $text, $variant ) {
 677                 global $wgDisableLangConversion;
 678                 if ( $wgDisableLangConversion ) {
 679                         return $text;
 680                 }
 681                 // Reset converter state for a new converter run.
 682                 $this->mConvRuleTitle = false;
 683                 return $this->recursiveConvertTopLevel( $text, $variant );
 684         }
 685
 686         /**
 687          * Recursively convert text on the outside. Allow to use nested
 688          * markups to custom rules.
 689          *
 690          * @param string $text Text to be converted
 691          * @param string $variant The target variant code
 692          * @param int $depth Depth of recursion
 693          * @return string Converted text
 694          */
 695         protected function recursiveConvertTopLevel( $text, $variant, $depth = 0 ) {
 696                 $startPos = 0;
 697                 $out = '';
 698                 $length = strlen( $text );
 699                 $shouldConvert = !$this->guessVariant( $text, $variant );
 700                 $continue = 1;
 701
 702                 $noScript = '<script.*?>.*?<\/script>(*SKIP)(*FAIL)';
 703                 $noStyle = '<style.*?>.*?<\/style>(*SKIP)(*FAIL)';
 704                 // phpcs:ignore Generic.Files.LineLength
 705                 $noHtml = '<(?:[^>=]*+(?>[^>=]*+=\s*+(?:"[^"]*"|\'[^\']*\'|[^\'">\s]*+))*+[^>=]*+>|.*+)(*SKIP)(*FAIL)';
 706                 while ( $startPos < $length && $continue ) {
 707                         $continue = preg_match(
 708                                 // Only match -{ outside of html.
 709                                 "/$noScript|$noStyle|$noHtml|-\{/",
 710                                 $text,
 711                                 $m,
 712                                 PREG_OFFSET_CAPTURE,
 713                                 $startPos
 714                         );
 715
 716                         if ( !$continue ) {
 717                                 // No more markup, append final segment
 718                                 $fragment = substr( $text, $startPos );
 719                                 $out .= $shouldConvert ? $this->autoConvert( $fragment, $variant ) : $fragment;
 720                                 return $out;
 721                         }
 722
 723                         // Offset of the match of the regex pattern.
 724                         $pos = $m[0][1];
 725
 726                         // Append initial segment
 727                         $fragment = substr( $text, $startPos, $pos - $startPos );
 728                         $out .= $shouldConvert ? $this->autoConvert( $fragment, $variant ) : $fragment;
 729                         // -{ marker found, not in attribute
 730                         // Advance position up to -{ marker.
 731                         $startPos = $pos;
 732                         // Do recursive conversion
 733                         // Note: This passes $startPos by reference, and advances it.
 734                         $out .= $this->recursiveConvertRule( $text, $variant, $startPos, $depth + 1 );
 735                 }
 736                 return $out;
 737         }
 738
 739         /**
 740          * Recursively convert text on the inside.
 741          *
 742          * @param string $text Text to be converted
 743          * @param string $variant The target variant code
 744          * @param int &$startPos
 745          * @param int $depth Depth of recursion
 746          *
 747          * @throws MWException
 748          * @return string Converted text
 749          */
 750         protected function recursiveConvertRule( $text, $variant, &$startPos, $depth = 0 ) {
 751                 // Quick sanity check (no function calls)
 752                 if ( $text[$startPos] !== '-' || $text[$startPos + 1] !== '{' ) {
 753                         throw new MWException( __METHOD__ . ': invalid input string' );
 754                 }
 755
 756                 $startPos += 2;
 757                 $inner = '';
 758                 $warningDone = false;
 759                 $length = strlen( $text );
 760
 761                 while ( $startPos < $length ) {
 762                         $m = false;
 763                         preg_match( '/-\{|\}-/', $text, $m, PREG_OFFSET_CAPTURE, $startPos );
 764                         if ( !$m ) {
 765                                 // Unclosed rule
 766                                 break;
 767                         }
 768
 769                         $token = $m[0][0];
 770                         $pos = $m[0][1];
 771
 772                         // Markup found
 773                         // Append initial segment
 774                         $inner .= substr( $text, $startPos, $pos - $startPos );
 775
 776                         // Advance position
 777                         $startPos = $pos;
 778
 779                         switch ( $token ) {
 780                                 case '-{':
 781                                         // Check max depth
 782                                         if ( $depth >= $this->mMaxDepth ) {
 783                                                 $inner .= '-{';
 784                                                 if ( !$warningDone ) {
 785                                                         $inner .= '<span class="error">' .
 786                                                                 wfMessage( 'language-converter-depth-warning' )
 787                                                                         ->numParams( $this->mMaxDepth )->inContentLanguage()->text() .
 788                                                                 '</span>';
 789                                                         $warningDone = true;
 790                                                 }
 791                                                 $startPos += 2;
 792                                                 continue;
 793                                         }
 794                                         // Recursively parse another rule
 795                                         $inner .= $this->recursiveConvertRule( $text, $variant, $startPos, $depth + 1 );
 796                                         break;
 797                                 case '}-':
 798                                         // Apply the rule
 799                                         $startPos += 2;
 800                                         $rule = new ConverterRule( $inner, $this );
 801                                         $rule->parse( $variant );
 802                                         $this->applyManualConv( $rule );
 803                                         return $rule->getDisplay();
 804                                 default:
 805                                         throw new MWException( __METHOD__ . ': invalid regex match' );
 806                         }
 807                 }
 808
 809                 // Unclosed rule
 810                 if ( $startPos < $length ) {
 811                         $inner .= substr( $text, $startPos );
 812                 }
 813                 $startPos = $length;
 814                 return '-{' . $this->autoConvert( $inner, $variant );
 815         }
 816
 817         /**
 818          * If a language supports multiple variants, it is possible that
 819          * non-existing link in one variant actually exists in another variant.
 820          * This function tries to find it. See e.g. LanguageZh.php
 821          * The input parameters may be modified upon return
 822          *
 823          * @param string &$link The name of the link
 824          * @param Title &$nt The title object of the link
 825          * @param bool $ignoreOtherCond To disable other conditions when
 826          *   we need to transclude a template or update a category's link
 827          */
 828         public function findVariantLink( &$link, &$nt, $ignoreOtherCond = false ) {
 829                 # If the article has already existed, there is no need to
 830                 # check it again, otherwise it may cause a fault.
 831                 if ( is_object( $nt ) && $nt->exists() ) {
 832                         return;
 833                 }
 834
 835                 global $wgDisableLangConversion, $wgDisableTitleConversion, $wgRequest;
 836                 $isredir = $wgRequest->getText( 'redirect', 'yes' );
 837                 $action = $wgRequest->getText( 'action' );
 838                 if ( $action == 'edit' && $wgRequest->getBool( 'redlink' ) ) {
 839                         $action = 'view';
 840                 }
 841                 $linkconvert = $wgRequest->getText( 'linkconvert', 'yes' );
 842                 $disableLinkConversion = $wgDisableLangConversion
 843                         || $wgDisableTitleConversion;
 844                 $linkBatch = new LinkBatch();
 845
 846                 $ns = NS_MAIN;
 847
 848                 if ( $disableLinkConversion ||
 849                         ( !$ignoreOtherCond &&
 850                                 ( $isredir == 'no'
 851                                         || $action == 'edit'
 852                                         || $action == 'submit'
 853                                         || $linkconvert == 'no' ) ) ) {
 854                         return;
 855                 }
 856
 857                 if ( is_object( $nt ) ) {
 858                         $ns = $nt->getNamespace();
 859                 }
 860
 861                 $variants = $this->autoConvertToAllVariants( $link );
 862                 if ( !$variants ) { // give up
 863                         return;
 864                 }
 865
 866                 $titles = [];
 867
 868                 foreach ( $variants as $v ) {
 869                         if ( $v != $link ) {
 870                                 $varnt = Title::newFromText( $v, $ns );
 871                                 if ( !is_null( $varnt ) ) {
 872                                         $linkBatch->addObj( $varnt );
 873                                         $titles[] = $varnt;
 874                                 }
 875                         }
 876                 }
 877
 878                 // fetch all variants in single query
 879                 $linkBatch->execute();
 880
 881                 foreach ( $titles as $varnt ) {
 882                         if ( $varnt->getArticleID() > 0 ) {
 883                                 $nt = $varnt;
 884                                 $link = $varnt->getText();
 885                                 break;
 886                         }
 887                 }
 888         }
 889
 890         /**
 891          * Returns language specific hash options.
 892          *
 893          * @return string
 894          */
 895         public function getExtraHashOptions() {
 896                 $variant = $this->getPreferredVariant();
 897
 898                 return '!' . $variant;
 899         }
 900
 901         /**
 902          * Guess if a text is written in a variant. This should be implemented in subclasses.
 903          *
 904          * @param string $text The text to be checked
 905          * @param string $variant Language code of the variant to be checked for
 906          * @return bool True if $text appears to be written in $variant, false if not
 907          *
 908          * @author Nikola Smolenski <smolensk@eunet.rs>
 909          * @since 1.19
 910          */
 911         public function guessVariant( $text, $variant ) {
 912                 return false;
 913         }
 914
 915         /**
 916          * Load default conversion tables.
 917          * This method must be implemented in derived class.
 918          *
 919          * @private
 920          * @throws MWException
 921          */
 922         function loadDefaultTables() {
 923                 $class = static::class;
 924                 throw new MWException( "Must implement loadDefaultTables() method in class $class" );
 925         }
 926
 927         /**
 928          * Load conversion tables either from the cache or the disk.
 929          * @private
 930          * @param bool $fromCache Load from memcached? Defaults to true.
 931          */
 932         function loadTables( $fromCache = true ) {
 933                 global $wgLanguageConverterCacheType;
 934
 935                 if ( $this->mTablesLoaded ) {
 936                         return;
 937                 }
 938
 939                 $this->mTablesLoaded = true;
 940                 $this->mTables = false;
 941                 $cache = ObjectCache::getInstance( $wgLanguageConverterCacheType );
 942                 $cacheKey = $cache->makeKey( 'conversiontables', $this->mMainLanguageCode );
 943                 if ( $fromCache ) {
 944                         $this->mTables = $cache->get( $cacheKey );
 945                 }
 946                 if ( !$this->mTables || !array_key_exists( self::CACHE_VERSION_KEY, $this->mTables ) ) {
 947                         // not in cache, or we need a fresh reload.
 948                         // We will first load the default tables
 949                         // then update them using things in MediaWiki:Conversiontable/*
 950                         $this->loadDefaultTables();
 951                         foreach ( $this->mVariants as $var ) {
 952                                 $cached = $this->parseCachedTable( $var );
 953                                 $this->mTables[$var]->mergeArray( $cached );
 954                         }
 955
 956                         $this->postLoadTables();
 957                         $this->mTables[self::CACHE_VERSION_KEY] = true;
 958
 959                         $cache->set( $cacheKey, $this->mTables, 43200 );
 960                 }
 961         }
 962
 963         /**
 964          * Hook for post processing after conversion tables are loaded.
 965          */
 966         function postLoadTables() {
 967         }
 968
 969         /**
 970          * Reload the conversion tables.
 971          *
 972          * Also used by test suites which need to reset the converter state.
 973          *
 974          * @private
 975          */
 976         private function reloadTables() {
 977                 if ( $this->mTables ) {
 978                         unset( $this->mTables );
 979                 }
 980
 981                 $this->mTablesLoaded = false;
 982                 $this->loadTables( false );
 983         }
 984
 985         /**
 986          * Parse the conversion table stored in the cache.
 987          *
 988          * The tables should be in blocks of the following form:
 989          *              -{
 990          *                      word => word ;
 991          *                      word => word ;
 992          *                      ...
 993          *              }-
 994          *
 995          * To make the tables more manageable, subpages are allowed
 996          * and will be parsed recursively if $recursive == true.
 997          *
 998          * @param string $code Language code
 999          * @param string $subpage Subpage name
1000          * @param bool $recursive Parse subpages recursively? Defaults to true.
1001          *
1002          * @return array
1003          */
1004         function parseCachedTable( $code, $subpage = '', $recursive = true ) {
1005                 static $parsed = [];
1006
1007                 $key = 'Conversiontable/' . $code;
1008                 if ( $subpage ) {
1009                         $key .= '/' . $subpage;
1010                 }
1011                 if ( array_key_exists( $key, $parsed ) ) {
1012                         return [];
1013                 }
1014
1015                 $parsed[$key] = true;
1016
1017                 if ( $subpage === '' ) {
1018                         $txt = MessageCache::singleton()->getMsgFromNamespace( $key, $code );
1019                 } else {
1020                         $txt = false;
1021                         $title = Title::makeTitleSafe( NS_MEDIAWIKI, $key );
1022                         if ( $title && $title->exists() ) {
1023                                 $revision = Revision::newFromTitle( $title );
1024                                 if ( $revision ) {
1025                                         if ( $revision->getContentModel() == CONTENT_MODEL_WIKITEXT ) {
1026                                                 $txt = $revision->getContent( Revision::RAW )->getNativeData();
1027                                         }
1028
1029                                         // @todo in the future, use a specialized content model, perhaps based on json!
1030                                 }
1031                         }
1032                 }
1033
1034                 # Nothing to parse if there's no text
1035                 if ( $txt === false || $txt === null || $txt === '' ) {
1036                         return [];
1037                 }
1038
1039                 // get all subpage links of the form
1040                 // [[MediaWiki:Conversiontable/zh-xx/...|...]]
1041                 $linkhead = $this->mLangObj->getNsText( NS_MEDIAWIKI ) .
1042                         ':Conversiontable';
1043                 $subs = StringUtils::explode( '[[', $txt );
1044                 $sublinks = [];
1045                 foreach ( $subs as $sub ) {
1046                         $link = explode( ']]', $sub, 2 );
1047                         if ( count( $link ) != 2 ) {
1048                                 continue;
1049                         }
1050                         $b = explode( '|', $link[0], 2 );
1051                         $b = explode( '/', trim( $b[0] ), 3 );
1052                         if ( count( $b ) == 3 ) {
1053                                 $sublink = $b[2];
1054                         } else {
1055                                 $sublink = '';
1056                         }
1057
1058                         if ( $b[0] == $linkhead && $b[1] == $code ) {
1059                                 $sublinks[] = $sublink;
1060                         }
1061                 }
1062
1063                 // parse the mappings in this page
1064                 $blocks = StringUtils::explode( '-{', $txt );
1065                 $ret = [];
1066                 $first = true;
1067                 foreach ( $blocks as $block ) {
1068                         if ( $first ) {
1069                                 // Skip the part before the first -{
1070                                 $first = false;
1071                                 continue;
1072                         }
1073                         $mappings = explode( '}-', $block, 2 )[0];
1074                         $stripped = str_replace( [ "'", '"', '*', '#' ], '', $mappings );
1075                         $table = StringUtils::explode( ';', $stripped );
1076                         foreach ( $table as $t ) {
1077                                 $m = explode( '=>', $t, 3 );
1078                                 if ( count( $m ) != 2 ) {
1079                                         continue;
1080                                 }
1081                                 // trim any trailling comments starting with '//'
1082                                 $tt = explode( '//', $m[1], 2 );
1083                                 $ret[trim( $m[0] )] = trim( $tt[0] );
1084                         }
1085                 }
1086
1087                 // recursively parse the subpages
1088                 if ( $recursive ) {
1089                         foreach ( $sublinks as $link ) {
1090                                 $s = $this->parseCachedTable( $code, $link, $recursive );
1091                                 $ret = $s + $ret;
1092                         }
1093                 }
1094
1095                 if ( $this->mUcfirst ) {
1096                         foreach ( $ret as $k => $v ) {
1097                                 $ret[$this->mLangObj->ucfirst( $k )] = $this->mLangObj->ucfirst( $v );
1098                         }
1099                 }
1100                 return $ret;
1101         }
1102
1103         /**
1104          * Enclose a string with the "no conversion" tag. This is used by
1105          * various functions in the Parser.
1106          *
1107          * @param string $text Text to be tagged for no conversion
1108          * @param bool $noParse Unused
1109          * @return string The tagged text
1110          */
1111         public function markNoConversion( $text, $noParse = false ) {
1112                 # don't mark if already marked
1113                 if ( strpos( $text, '-{' ) || strpos( $text, '}-' ) ) {
1114                         return $text;
1115                 }
1116
1117                 $ret = "-{R|$text}-";
1118                 return $ret;
1119         }
1120
1121         /**
1122          * Convert the sorting key for category links. This should make different
1123          * keys that are variants of each other map to the same key.
1124          *
1125          * @param string $key
1126          *
1127          * @return string
1128          */
1129         function convertCategoryKey( $key ) {
1130                 return $key;
1131         }
1132
1133         /**
1134          * Refresh the cache of conversion tables when
1135          * MediaWiki:Conversiontable* is updated.
1136          *
1137          * @param Title $titleobj The Title of the page being updated
1138          */
1139         public function updateConversionTable( Title $titleobj ) {
1140                 if ( $titleobj->getNamespace() == NS_MEDIAWIKI ) {
1141                         $title = $titleobj->getDBkey();
1142                         $t = explode( '/', $title, 3 );
1143                         $c = count( $t );
1144                         if ( $c > 1 && $t[0] == 'Conversiontable' ) {
1145                                 if ( $this->validateVariant( $t[1] ) ) {
1146                                         $this->reloadTables();
1147                                 }
1148                         }
1149                 }
1150         }
1151
1152         /**
1153          * Get the cached separator pattern for ConverterRule::parseRules()
1154          * @return string
1155          */
1156         function getVarSeparatorPattern() {
1157                 if ( is_null( $this->mVarSeparatorPattern ) ) {
1158                         // varsep_pattern for preg_split:
1159                         // text should be splited by ";" only if a valid variant
1160                         // name exist after the markup, for example:
1161                         //  -{zh-hans:<span style="font-size:120%;">xxx</span>;zh-hant:\
1162                         //  <span style="font-size:120%;">yyy</span>;}-
1163                         // we should split it as:
1164                         //  [
1165                         //    [0] => 'zh-hans:<span style="font-size:120%;">xxx</span>'
1166                         //    [1] => 'zh-hant:<span style="font-size:120%;">yyy</span>'
1167                         //    [2] => ''
1168                         //  ]
1169                         $pat = '/;\s*(?=';
1170                         foreach ( $this->mVariants as $variant ) {
1171                                 // zh-hans:xxx;zh-hant:yyy
1172                                 $pat .= $variant . '\s*:|';
1173                                 // xxx=>zh-hans:yyy; xxx=>zh-hant:zzz
1174                                 $pat .= '[^;]*?=>\s*' . $variant . '\s*:|';
1175                         }
1176                         $pat .= '\s*$)/';
1177                         $this->mVarSeparatorPattern = $pat;
1178                 }
1179                 return $this->mVarSeparatorPattern;
1180         }
1181 }