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
  22 /**
  23  * Base class for language conversion.
  24  * @ingroup Language
  25  *
  26  * @author Zhengzhu Feng <zhengzhu@gmail.com>
  27  * @maintainers fdcn <fdcn64@gmail.com>, shinjiman <shinjiman@gmail.com>, PhiLiP <philip.npc@gmail.com>
  28  */
  29 class LanguageConverter {
  30
  31         /**
  32          * languages supporting variants
  33          * @since 1.20
  34          * @var array
  35          */
  36         static public $languagesWithVariants = array(
  37                 'gan',
  38                 'iu',
  39                 'kk',
  40                 'ku',
  41                 'shi',
  42                 'sr',
  43                 'tg',
  44                 'uz',
  45                 'zh',
  46         );
  47
  48         public $mMainLanguageCode;
  49         public $mVariants, $mVariantFallbacks, $mVariantNames;
  50         public $mTablesLoaded = false;
  51         public $mTables;
  52         // 'bidirectional' 'unidirectional' 'disable' for each variant
  53         public $mManualLevel;
  54
  55         /**
  56          * @var string memcached key name
  57          */
  58         public $mCacheKey;
  59
  60         public $mLangObj;
  61         public $mFlags;
  62         public $mDescCodeSep = ':', $mDescVarSep = ';';
  63         public $mUcfirst = false;
  64         public $mConvRuleTitle = false;
  65         public $mURLVariant;
  66         public $mUserVariant;
  67         public $mHeaderVariant;
  68         public $mMaxDepth = 10;
  69         public $mVarSeparatorPattern;
  70
  71         const CACHE_VERSION_KEY = 'VERSION 7';
  72
  73         /**
  74          * Constructor
  75          *
  76          * @param Language $langobj
  77          * @param string $maincode The main language code of this language
  78          * @param array $variants The supported variants of this language
  79          * @param array $variantfallbacks The fallback language of each variant
  80          * @param array $flags Defining the custom strings that maps to the flags
  81          * @param array $manualLevel Limit for supported variants
  82          */
  83         public function __construct( $langobj, $maincode, $variants = array(),
  84                                                                 $variantfallbacks = array(), $flags = array(),
  85                                                                 $manualLevel = array() ) {
  86                 global $wgDisabledVariants;
  87                 $this->mLangObj = $langobj;
  88                 $this->mMainLanguageCode = $maincode;
  89                 $this->mVariants = array_diff( $variants, $wgDisabledVariants );
  90                 $this->mVariantFallbacks = $variantfallbacks;
  91                 $this->mVariantNames = Language::fetchLanguageNames();
  92                 $this->mCacheKey = wfMemcKey( 'conversiontables', $maincode );
  93                 $defaultflags = array(
  94                         // 'S' show converted text
  95                         // '+' add rules for alltext
  96                         // 'E' the gave flags is error
  97                         // these flags above are reserved for program
  98                         'A' => 'A',       // add rule for convert code (all text convert)
  99                         'T' => 'T',       // title convert
 100                         'R' => 'R',       // raw content
 101                         'D' => 'D',       // convert description (subclass implement)
 102                         '-' => '-',       // remove convert (not implement)
 103                         'H' => 'H',       // add rule for convert code (but no display in placed code)
 104                         'N' => 'N'        // current variant name
 105                 );
 106                 $this->mFlags = array_merge( $defaultflags, $flags );
 107                 foreach ( $this->mVariants as $v ) {
 108                         if ( array_key_exists( $v, $manualLevel ) ) {
 109                                 $this->mManualLevel[$v] = $manualLevel[$v];
 110                         } else {
 111                                 $this->mManualLevel[$v] = 'bidirectional';
 112                         }
 113                         $this->mFlags[$v] = $v;
 114                 }
 115         }
 116
 117         /**
 118          * Get all valid variants.
 119          * Call this instead of using $this->mVariants directly.
 120          *
 121          * @return array Contains all valid variants
 122          */
 123         public function getVariants() {
 124                 return $this->mVariants;
 125         }
 126
 127         /**
 128          * In case some variant is not defined in the markup, we need
 129          * to have some fallback. For example, in zh, normally people
 130          * will define zh-hans and zh-hant, but less so for zh-sg or zh-hk.
 131          * when zh-sg is preferred but not defined, we will pick zh-hans
 132          * in this case. Right now this is only used by zh.
 133          *
 134          * @param string $variant The language code of the variant
 135          * @return string|array The code of the fallback language or the
 136          *   main code if there is no fallback
 137          */
 138         public function getVariantFallbacks( $variant ) {
 139                 if ( isset( $this->mVariantFallbacks[$variant] ) ) {
 140                         return $this->mVariantFallbacks[$variant];
 141                 }
 142                 return $this->mMainLanguageCode;
 143         }
 144
 145         /**
 146          * Get the title produced by the conversion rule.
 147          * @return string The converted title text
 148          */
 149         public function getConvRuleTitle() {
 150                 return $this->mConvRuleTitle;
 151         }
 152
 153         /**
 154          * Get preferred language variant.
 155          * @return string The preferred language code
 156          */
 157         public function getPreferredVariant() {
 158                 global $wgDefaultLanguageVariant, $wgUser;
 159
 160                 $req = $this->getURLVariant();
 161
 162                 if ( $wgUser->isLoggedIn() && !$req ) {
 163                         $req = $this->getUserVariant();
 164                 } elseif ( !$req ) {
 165                         $req = $this->getHeaderVariant();
 166                 }
 167
 168                 if ( $wgDefaultLanguageVariant && !$req ) {
 169                         $req = $this->validateVariant( $wgDefaultLanguageVariant );
 170                 }
 171
 172                 // This function, unlike the other get*Variant functions, is
 173                 // not memoized (i.e. there return value is not cached) since
 174                 // new information might appear during processing after this
 175                 // is first called.
 176                 if ( $this->validateVariant( $req ) ) {
 177                         return $req;
 178                 }
 179                 return $this->mMainLanguageCode;
 180         }
 181
 182         /**
 183          * Get default variant.
 184          * This function would not be affected by user's settings
 185          * @return string The default variant code
 186          */
 187         public function getDefaultVariant() {
 188                 global $wgDefaultLanguageVariant;
 189
 190                 $req = $this->getURLVariant();
 191
 192                 if ( !$req ) {
 193                         $req = $this->getHeaderVariant();
 194                 }
 195
 196                 if ( $wgDefaultLanguageVariant && !$req ) {
 197                         $req = $this->validateVariant( $wgDefaultLanguageVariant );
 198                 }
 199
 200                 if ( $req ) {
 201                         return $req;
 202                 }
 203                 return $this->mMainLanguageCode;
 204         }
 205
 206         /**
 207          * Validate the variant
 208          * @param string $variant The variant to validate
 209          * @return mixed Returns the variant if it is valid, null otherwise
 210          */
 211         public function validateVariant( $variant = null ) {
 212                 if ( $variant !== null && in_array( $variant, $this->mVariants ) ) {
 213                         return $variant;
 214                 }
 215                 return null;
 216         }
 217
 218         /**
 219          * Get the variant specified in the URL
 220          *
 221          * @return mixed Variant if one found, false otherwise.
 222          */
 223         public function getURLVariant() {
 224                 global $wgRequest;
 225
 226                 if ( $this->mURLVariant ) {
 227                         return $this->mURLVariant;
 228                 }
 229
 230                 // see if the preference is set in the request
 231                 $ret = $wgRequest->getText( 'variant' );
 232
 233                 if ( !$ret ) {
 234                         $ret = $wgRequest->getVal( 'uselang' );
 235                 }
 236
 237                 $this->mURLVariant = $this->validateVariant( $ret );
 238                 return $this->mURLVariant;
 239         }
 240
 241         /**
 242          * Determine if the user has a variant set.
 243          *
 244          * @return mixed Variant if one found, false otherwise.
 245          */
 246         protected function getUserVariant() {
 247                 global $wgUser, $wgContLang;
 248
 249                 // memoizing this function wreaks havoc on parserTest.php
 250                 /*
 251                 if ( $this->mUserVariant ) {
 252                         return $this->mUserVariant;
 253                 }
 254                 */
 255
 256                 // Get language variant preference from logged in users
 257                 // Don't call this on stub objects because that causes infinite
 258                 // recursion during initialisation
 259                 if ( $wgUser->isLoggedIn() ) {
 260                         if ( $this->mMainLanguageCode == $wgContLang->getCode() ) {
 261                                 $ret = $wgUser->getOption( 'variant' );
 262                         } else {
 263                                 $ret = $wgUser->getOption( 'variant-' . $this->mMainLanguageCode );
 264                         }
 265                 } else {
 266                         // figure out user lang without constructing wgLang to avoid
 267                         // infinite recursion
 268                         $ret = $wgUser->getOption( 'language' );
 269                 }
 270
 271                 $this->mUserVariant = $this->validateVariant( $ret );
 272                 return $this->mUserVariant;
 273         }
 274
 275         /**
 276          * Determine the language variant from the Accept-Language header.
 277          *
 278          * @return mixed Variant if one found, false otherwise.
 279          */
 280         protected function getHeaderVariant() {
 281                 global $wgRequest;
 282
 283                 if ( $this->mHeaderVariant ) {
 284                         return $this->mHeaderVariant;
 285                 }
 286
 287                 // see if some supported language variant is set in the
 288                 // HTTP header.
 289                 $languages = array_keys( $wgRequest->getAcceptLang() );
 290                 if ( empty( $languages ) ) {
 291                         return null;
 292                 }
 293
 294                 $fallbackLanguages = array();
 295                 foreach ( $languages as $language ) {
 296                         $this->mHeaderVariant = $this->validateVariant( $language );
 297                         if ( $this->mHeaderVariant ) {
 298                                 break;
 299                         }
 300
 301                         // To see if there are fallbacks of current language.
 302                         // We record these fallback variants, and process
 303                         // them later.
 304                         $fallbacks = $this->getVariantFallbacks( $language );
 305                         if ( is_string( $fallbacks ) && $fallbacks !== $this->mMainLanguageCode ) {
 306                                 $fallbackLanguages[] = $fallbacks;
 307                         } elseif ( is_array( $fallbacks ) ) {
 308                                 $fallbackLanguages =
 309                                         array_merge( $fallbackLanguages, $fallbacks );
 310                         }
 311                 }
 312
 313                 if ( !$this->mHeaderVariant ) {
 314                         // process fallback languages now
 315                         $fallback_languages = array_unique( $fallbackLanguages );
 316                         foreach ( $fallback_languages as $language ) {
 317                                 $this->mHeaderVariant = $this->validateVariant( $language );
 318                                 if ( $this->mHeaderVariant ) {
 319                                         break;
 320                                 }
 321                         }
 322                 }
 323
 324                 return $this->mHeaderVariant;
 325         }
 326
 327         /**
 328          * Dictionary-based conversion.
 329          * This function would not parse the conversion rules.
 330          * If you want to parse rules, try to use convert() or
 331          * convertTo().
 332          *
 333          * @param $text String the text to be converted
 334          * @param $toVariant bool|string the target language code
 335          * @return String the converted text
 336          */
 337         public function autoConvert( $text, $toVariant = false ) {
 338                 wfProfileIn( __METHOD__ );
 339
 340                 $this->loadTables();
 341
 342                 if ( !$toVariant ) {
 343                         $toVariant = $this->getPreferredVariant();
 344                         if ( !$toVariant ) {
 345                                 wfProfileOut( __METHOD__ );
 346                                 return $text;
 347                         }
 348                 }
 349
 350                 if ( $this->guessVariant( $text, $toVariant ) ) {
 351                         wfProfileOut( __METHOD__ );
 352                         return $text;
 353                 }
 354
 355                 /* we convert everything except:
 356                    1. HTML markups (anything between < and >)
 357                    2. HTML entities
 358                    3. placeholders created by the parser
 359                 */
 360                 global $wgParser;
 361                 if ( isset( $wgParser ) && $wgParser->UniqPrefix() != '' ) {
 362                         $marker = '|' . $wgParser->UniqPrefix() . '[\-a-zA-Z0-9]+';
 363                 } else {
 364                         $marker = '';
 365                 }
 366
 367                 // this one is needed when the text is inside an HTML markup
 368                 $htmlfix = '|<[^>]+$|^[^<>]*>';
 369
 370                 // disable convert to variants between <code> tags
 371                 $codefix = '<code>.+?<\/code>|';
 372                 // disable conversion of <script> tags
 373                 $scriptfix = '<script.*?>.*?<\/script>|';
 374                 // disable conversion of <pre> tags
 375                 $prefix = '<pre.*?>.*?<\/pre>|';
 376
 377                 $reg = '/' . $codefix . $scriptfix . $prefix .
 378                         '<[^>]+>|&[a-zA-Z#][a-z0-9]+;' . $marker . $htmlfix . '/s';
 379                 $startPos = 0;
 380                 $sourceBlob = '';
 381                 $literalBlob = '';
 382
 383                 // Guard against delimiter nulls in the input
 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 ( array( '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                                         // Remove HTML tags to avoid disrupting the layout
 420                                         $attr = preg_replace( '/<[^>]+>/', '', $attr );
 421                                         if ( $attr !== $attrs[$attrName] ) {
 422                                                 $attrs[$attrName] = $attr;
 423                                                 $changed = true;
 424                                         }
 425                                 }
 426                                 if ( $changed ) {
 427                                         $element = $elementMatches[1] . Html::expandAttributes( $attrs ) .
 428                                                 $elementMatches[3];
 429                                 }
 430                         }
 431                         $literalBlob .= $element . "\000";
 432                 }
 433
 434                 // Do the main translation batch
 435                 $translatedBlob = $this->translate( $sourceBlob, $toVariant );
 436
 437                 // Put the output back together
 438                 $translatedIter = StringUtils::explode( "\000", $translatedBlob );
 439                 $literalIter = StringUtils::explode( "\000", $literalBlob );
 440                 $output = '';
 441                 while ( $translatedIter->valid() && $literalIter->valid() ) {
 442                         $output .= $translatedIter->current();
 443                         $output .= $literalIter->current();
 444                         $translatedIter->next();
 445                         $literalIter->next();
 446                 }
 447
 448                 wfProfileOut( __METHOD__ );
 449                 return $output;
 450         }
 451
 452         /**
 453          * Translate a string to a variant.
 454          * Doesn't parse rules or do any of that other stuff, for that use
 455          * convert() or convertTo().
 456          *
 457          * @param string $text Text to convert
 458          * @param string $variant Variant language code
 459          * @return string Translated text
 460          */
 461         public function translate( $text, $variant ) {
 462                 wfProfileIn( __METHOD__ );
 463                 // If $text is empty or only includes spaces, do nothing
 464                 // Otherwise translate it
 465                 if ( trim( $text ) ) {
 466                         $this->loadTables();
 467                         $text = $this->mTables[$variant]->replace( $text );
 468                 }
 469                 wfProfileOut( __METHOD__ );
 470                 return $text;
 471         }
 472
 473         /**
 474          * Call translate() to convert text to all valid variants.
 475          *
 476          * @param string $text The text to be converted
 477          * @return array Variant => converted text
 478          */
 479         public function autoConvertToAllVariants( $text ) {
 480                 wfProfileIn( __METHOD__ );
 481                 $this->loadTables();
 482
 483                 $ret = array();
 484                 foreach ( $this->mVariants as $variant ) {
 485                         $ret[$variant] = $this->translate( $text, $variant );
 486                 }
 487
 488                 wfProfileOut( __METHOD__ );
 489                 return $ret;
 490         }
 491
 492         /**
 493          * Apply manual conversion rules.
 494          *
 495          * @param $convRule ConverterRule Object of ConverterRule
 496          */
 497         protected function applyManualConv( $convRule ) {
 498                 // Use syntax -{T|zh-cn:TitleCN; zh-tw:TitleTw}- to custom
 499                 // title conversion.
 500                 // Bug 24072: $mConvRuleTitle was overwritten by other manual
 501                 // rule(s) not for title, this breaks the title conversion.
 502                 $newConvRuleTitle = $convRule->getTitle();
 503                 if ( $newConvRuleTitle ) {
 504                         // So I add an empty check for getTitle()
 505                         $this->mConvRuleTitle = $newConvRuleTitle;
 506                 }
 507
 508                 // merge/remove manual conversion rules to/from global table
 509                 $convTable = $convRule->getConvTable();
 510                 $action = $convRule->getRulesAction();
 511                 foreach ( $convTable as $variant => $pair ) {
 512                         if ( !$this->validateVariant( $variant ) ) {
 513                                 continue;
 514                         }
 515
 516                         if ( $action == 'add' ) {
 517                                 foreach ( $pair as $from => $to ) {
 518                                         // to ensure that $from and $to not be left blank
 519                                         // so $this->translate() could always return a string
 520                                         if ( $from || $to ) {
 521                                                 // more efficient than array_merge(), about 2.5 times.
 522                                                 $this->mTables[$variant]->setPair( $from, $to );
 523                                         }
 524                                 }
 525                         } elseif ( $action == 'remove' ) {
 526                                 $this->mTables[$variant]->removeArray( $pair );
 527                         }
 528                 }
 529         }
 530
 531         /**
 532          * Auto convert a Title object to a readable string in the
 533          * preferred variant.
 534          *
 535          * @param Title $title A object of Title
 536          * @return string Converted title text
 537          */
 538         public function convertTitle( $title ) {
 539                 $variant = $this->getPreferredVariant();
 540                 $index = $title->getNamespace();
 541                 if ( $index !== NS_MAIN ) {
 542                         $text = $this->convertNamespace( $index, $variant ) . ':';
 543                 } else {
 544                         $text = '';
 545                 }
 546                 $text .= $this->translate( $title->getText(), $variant );
 547                 return $text;
 548         }
 549
 550         /**
 551          * Get the namespace display name in the preferred variant.
 552          *
 553          * @param int $index Namespace id
 554          * @param string|null $variant Variant code or null for preferred variant
 555          * @return string Namespace name for display
 556          */
 557         public function convertNamespace( $index, $variant = null ) {
 558                 if ( $variant === null ) {
 559                         $variant = $this->getPreferredVariant();
 560                 }
 561                 if ( $index === NS_MAIN ) {
 562                         return '';
 563                 } else {
 564                         // First check if a message gives a converted name in the target variant.
 565                         $nsConvMsg = wfMessage( 'conversion-ns' . $index )->inLanguage( $variant );
 566                         if ( $nsConvMsg->exists() ) {
 567                                 return $nsConvMsg->plain();
 568                         }
 569                         // Then check if a message gives a converted name in content language
 570                         // which needs extra translation to the target variant.
 571                         $nsConvMsg = wfMessage( 'conversion-ns' . $index )->inContentLanguage();
 572                         if ( $nsConvMsg->exists() ) {
 573                                 return $this->translate( $nsConvMsg->plain(), $variant );
 574                         }
 575                         // No message exists, retrieve it from the target variant's namespace names.
 576                         $langObj = $this->mLangObj->factory( $variant );
 577                         return $langObj->getFormattedNsText( $index );
 578                 }
 579         }
 580
 581         /**
 582          * Convert text to different variants of a language. The automatic
 583          * conversion is done in autoConvert(). Here we parse the text
 584          * marked with -{}-, which specifies special conversions of the
 585          * text that can not be accomplished in autoConvert().
 586          *
 587          * Syntax of the markup:
 588          * -{code1:text1;code2:text2;...}-  or
 589          * -{flags|code1:text1;code2:text2;...}-  or
 590          * -{text}- in which case no conversion should take place for text
 591          *
 592          * @param string $text Text to be converted
 593          * @return string Converted text
 594          */
 595         public function convert( $text ) {
 596                 $variant = $this->getPreferredVariant();
 597                 return $this->convertTo( $text, $variant );
 598         }
 599
 600         /**
 601          * Same as convert() except a extra parameter to custom variant.
 602          *
 603          * @param string $text Text to be converted
 604          * @param string $variant The target variant code
 605          * @return string Converted text
 606          */
 607         public function convertTo( $text, $variant ) {
 608                 global $wgDisableLangConversion;
 609                 if ( $wgDisableLangConversion ) {
 610                         return $text;
 611                 }
 612                 // Reset converter state for a new converter run.
 613                 $this->mConvRuleTitle = false;
 614                 return $this->recursiveConvertTopLevel( $text, $variant );
 615         }
 616
 617         /**
 618          * Recursively convert text on the outside. Allow to use nested
 619          * markups to custom rules.
 620          *
 621          * @param string $text Text to be converted
 622          * @param string $variant The target variant code
 623          * @param int $depth Depth of recursion
 624          * @return string Converted text
 625          */
 626         protected function recursiveConvertTopLevel( $text, $variant, $depth = 0 ) {
 627                 $startPos = 0;
 628                 $out = '';
 629                 $length = strlen( $text );
 630                 $shouldConvert = !$this->guessVariant( $text, $variant );
 631
 632                 while ( $startPos < $length ) {
 633                         $pos = strpos( $text, '-{', $startPos );
 634
 635                         if ( $pos === false ) {
 636                                 // No more markup, append final segment
 637                                 $fragment = substr( $text, $startPos );
 638                                 $out .= $shouldConvert ? $this->autoConvert( $fragment, $variant ) : $fragment;
 639                                 return $out;
 640                         }
 641
 642                         // Markup found
 643                         // Append initial segment
 644                         $fragment = substr( $text, $startPos, $pos - $startPos );
 645                         $out .= $shouldConvert ? $this->autoConvert( $fragment, $variant ) : $fragment;
 646
 647                         // Advance position
 648                         $startPos = $pos;
 649
 650                         // Do recursive conversion
 651                         $out .= $this->recursiveConvertRule( $text, $variant, $startPos, $depth + 1 );
 652                 }
 653
 654                 return $out;
 655         }
 656
 657         /**
 658          * Recursively convert text on the inside.
 659          *
 660          * @param string $text Text to be converted
 661          * @param string $variant The target variant code
 662          * @param int $startPos
 663          * @param int $depth Depth of recursion
 664          *
 665          * @throws MWException
 666          * @return string Converted text
 667          */
 668         protected function recursiveConvertRule( $text, $variant, &$startPos, $depth = 0 ) {
 669                 // Quick sanity check (no function calls)
 670                 if ( $text[$startPos] !== '-' || $text[$startPos + 1] !== '{' ) {
 671                         throw new MWException( __METHOD__ . ': invalid input string' );
 672                 }
 673
 674                 $startPos += 2;
 675                 $inner = '';
 676                 $warningDone = false;
 677                 $length = strlen( $text );
 678
 679                 while ( $startPos < $length ) {
 680                         $m = false;
 681                         preg_match( '/-\{|\}-/', $text, $m, PREG_OFFSET_CAPTURE, $startPos );
 682                         if ( !$m ) {
 683                                 // Unclosed rule
 684                                 break;
 685                         }
 686
 687                         $token = $m[0][0];
 688                         $pos = $m[0][1];
 689
 690                         // Markup found
 691                         // Append initial segment
 692                         $inner .= substr( $text, $startPos, $pos - $startPos );
 693
 694                         // Advance position
 695                         $startPos = $pos;
 696
 697                         switch ( $token ) {
 698                                 case '-{':
 699                                         // Check max depth
 700                                         if ( $depth >= $this->mMaxDepth ) {
 701                                                 $inner .= '-{';
 702                                                 if ( !$warningDone ) {
 703                                                         $inner .= '<span class="error">' .
 704                                                                 wfMessage( 'language-converter-depth-warning' )
 705                                                                         ->numParams( $this->mMaxDepth )->inContentLanguage()->text() .
 706                                                                 '</span>';
 707                                                         $warningDone = true;
 708                                                 }
 709                                                 $startPos += 2;
 710                                                 continue;
 711                                         }
 712                                         // Recursively parse another rule
 713                                         $inner .= $this->recursiveConvertRule( $text, $variant, $startPos, $depth + 1 );
 714                                         break;
 715                                 case '}-':
 716                                         // Apply the rule
 717                                         $startPos += 2;
 718                                         $rule = new ConverterRule( $inner, $this );
 719                                         $rule->parse( $variant );
 720                                         $this->applyManualConv( $rule );
 721                                         return $rule->getDisplay();
 722                                 default:
 723                                         throw new MWException( __METHOD__ . ': invalid regex match' );
 724                         }
 725                 }
 726
 727                 // Unclosed rule
 728                 if ( $startPos < $length ) {
 729                         $inner .= substr( $text, $startPos );
 730                 }
 731                 $startPos = $length;
 732                 return '-{' . $this->autoConvert( $inner, $variant );
 733         }
 734
 735         /**
 736          * If a language supports multiple variants, it is possible that
 737          * non-existing link in one variant actually exists in another variant.
 738          * This function tries to find it. See e.g. LanguageZh.php
 739          *
 740          * @param string $link The name of the link
 741          * @param mixed $nt The title object of the link
 742          * @param bool $ignoreOtherCond To disable other conditions when
 743          *   we need to transclude a template or update a category's link
 744          * @return void Null, the input parameters may be modified upon return
 745          */
 746         public function findVariantLink( &$link, &$nt, $ignoreOtherCond = false ) {
 747                 # If the article has already existed, there is no need to
 748                 # check it again, otherwise it may cause a fault.
 749                 if ( is_object( $nt ) && $nt->exists() ) {
 750                         return;
 751                 }
 752
 753                 global $wgDisableLangConversion, $wgDisableTitleConversion, $wgRequest;
 754                 $isredir = $wgRequest->getText( 'redirect', 'yes' );
 755                 $action = $wgRequest->getText( 'action' );
 756                 $linkconvert = $wgRequest->getText( 'linkconvert', 'yes' );
 757                 $disableLinkConversion = $wgDisableLangConversion
 758                         || $wgDisableTitleConversion;
 759                 $linkBatch = new LinkBatch();
 760
 761                 $ns = NS_MAIN;
 762
 763                 if ( $disableLinkConversion ||
 764                         ( !$ignoreOtherCond &&
 765                                 ( $isredir == 'no'
 766                                         || $action == 'edit'
 767                                         || $action == 'submit'
 768                                         || $linkconvert == 'no' ) ) ) {
 769                         return;
 770                 }
 771
 772                 if ( is_object( $nt ) ) {
 773                         $ns = $nt->getNamespace();
 774                 }
 775
 776                 $variants = $this->autoConvertToAllVariants( $link );
 777                 if ( !$variants ) { // give up
 778                         return;
 779                 }
 780
 781                 $titles = array();
 782
 783                 foreach ( $variants as $v ) {
 784                         if ( $v != $link ) {
 785                                 $varnt = Title::newFromText( $v, $ns );
 786                                 if ( !is_null( $varnt ) ) {
 787                                         $linkBatch->addObj( $varnt );
 788                                         $titles[] = $varnt;
 789                                 }
 790                         }
 791                 }
 792
 793                 // fetch all variants in single query
 794                 $linkBatch->execute();
 795
 796                 foreach ( $titles as $varnt ) {
 797                         if ( $varnt->getArticleID() > 0 ) {
 798                                 $nt = $varnt;
 799                                 $link = $varnt->getText();
 800                                 break;
 801                         }
 802                 }
 803         }
 804
 805         /**
 806          * Returns language specific hash options.
 807          *
 808          * @return string
 809          */
 810         public function getExtraHashOptions() {
 811                 $variant = $this->getPreferredVariant();
 812                 return '!' . $variant;
 813         }
 814
 815         /**
 816          * Guess if a text is written in a variant. This should be implemented in subclasses.
 817          *
 818          * @param string $text the text to be checked
 819          * @param string $variant language code of the variant to be checked for
 820          * @return bool true if $text appears to be written in $variant, false if not
 821          *
 822          * @author Nikola Smolenski <smolensk@eunet.rs>
 823          * @since 1.19
 824          */
 825         public function guessVariant( $text, $variant ) {
 826                 return false;
 827         }
 828
 829         /**
 830          * Load default conversion tables.
 831          * This method must be implemented in derived class.
 832          *
 833          * @private
 834          * @throws MWException
 835          */
 836         function loadDefaultTables() {
 837                 $name = get_class( $this );
 838                 throw new MWException( "Must implement loadDefaultTables() method in class $name" );
 839         }
 840
 841         /**
 842          * Load conversion tables either from the cache or the disk.
 843          * @private
 844          * @param bool $fromCache Load from memcached? Defaults to true.
 845          */
 846         function loadTables( $fromCache = true ) {
 847                 global $wgLangConvMemc;
 848
 849                 if ( $this->mTablesLoaded ) {
 850                         return;
 851                 }
 852
 853                 wfProfileIn( __METHOD__ );
 854                 $this->mTablesLoaded = true;
 855                 $this->mTables = false;
 856                 if ( $fromCache ) {
 857                         wfProfileIn( __METHOD__ . '-cache' );
 858                         $this->mTables = $wgLangConvMemc->get( $this->mCacheKey );
 859                         wfProfileOut( __METHOD__ . '-cache' );
 860                 }
 861                 if ( !$this->mTables || !array_key_exists( self::CACHE_VERSION_KEY, $this->mTables ) ) {
 862                         wfProfileIn( __METHOD__ . '-recache' );
 863                         // not in cache, or we need a fresh reload.
 864                         // We will first load the default tables
 865                         // then update them using things in MediaWiki:Conversiontable/*
 866                         $this->loadDefaultTables();
 867                         foreach ( $this->mVariants as $var ) {
 868                                 $cached = $this->parseCachedTable( $var );
 869                                 $this->mTables[$var]->mergeArray( $cached );
 870                         }
 871
 872                         $this->postLoadTables();
 873                         $this->mTables[self::CACHE_VERSION_KEY] = true;
 874
 875                         $wgLangConvMemc->set( $this->mCacheKey, $this->mTables, 43200 );
 876                         wfProfileOut( __METHOD__ . '-recache' );
 877                 }
 878                 wfProfileOut( __METHOD__ );
 879         }
 880
 881         /**
 882          * Hook for post processing after conversion tables are loaded.
 883          */
 884         function postLoadTables() { }
 885
 886         /**
 887          * Reload the conversion tables.
 888          *
 889          * @private
 890          */
 891         function reloadTables() {
 892                 if ( $this->mTables ) {
 893                         unset( $this->mTables );
 894                 }
 895                 $this->mTablesLoaded = false;
 896                 $this->loadTables( false );
 897         }
 898
 899         /**
 900          * Parse the conversion table stored in the cache.
 901          *
 902          * The tables should be in blocks of the following form:
 903          *              -{
 904          *                      word => word ;
 905          *                      word => word ;
 906          *                      ...
 907          *              }-
 908          *
 909          * To make the tables more manageable, subpages are allowed
 910          * and will be parsed recursively if $recursive == true.
 911          *
 912          * @param string $code Language code
 913          * @param string $subpage Subpage name
 914          * @param bool $recursive Parse subpages recursively? Defaults to true.
 915          *
 916          * @return array
 917          */
 918         function parseCachedTable( $code, $subpage = '', $recursive = true ) {
 919                 static $parsed = array();
 920
 921                 $key = 'Conversiontable/' . $code;
 922                 if ( $subpage ) {
 923                         $key .= '/' . $subpage;
 924                 }
 925                 if ( array_key_exists( $key, $parsed ) ) {
 926                         return array();
 927                 }
 928
 929                 $parsed[$key] = true;
 930
 931                 if ( $subpage === '' ) {
 932                         $txt = MessageCache::singleton()->getMsgFromNamespace( $key, $code );
 933                 } else {
 934                         $txt = false;
 935                         $title = Title::makeTitleSafe( NS_MEDIAWIKI, $key );
 936                         if ( $title && $title->exists() ) {
 937                                 $revision = Revision::newFromTitle( $title );
 938                                 if ( $revision ) {
 939                                         if ( $revision->getContentModel() == CONTENT_MODEL_WIKITEXT ) {
 940                                                 $txt = $revision->getContent( Revision::RAW )->getNativeData();
 941                                         }
 942
 943                                         // @todo in the future, use a specialized content model, perhaps based on json!
 944                                 }
 945                         }
 946                 }
 947
 948                 # Nothing to parse if there's no text
 949                 if ( $txt === false || $txt === null || $txt === '' ) {
 950                         return array();
 951                 }
 952
 953                 // get all subpage links of the form
 954                 // [[MediaWiki:Conversiontable/zh-xx/...|...]]
 955                 $linkhead = $this->mLangObj->getNsText( NS_MEDIAWIKI ) .
 956                         ':Conversiontable';
 957                 $subs = StringUtils::explode( '[[', $txt );
 958                 $sublinks = array();
 959                 foreach ( $subs as $sub ) {
 960                         $link = explode( ']]', $sub, 2 );
 961                         if ( count( $link ) != 2 ) {
 962                                 continue;
 963                         }
 964                         $b = explode( '|', $link[0], 2 );
 965                         $b = explode( '/', trim( $b[0] ), 3 );
 966                         if ( count( $b ) == 3 ) {
 967                                 $sublink = $b[2];
 968                         } else {
 969                                 $sublink = '';
 970                         }
 971
 972                         if ( $b[0] == $linkhead && $b[1] == $code ) {
 973                                 $sublinks[] = $sublink;
 974                         }
 975                 }
 976
 977                 // parse the mappings in this page
 978                 $blocks = StringUtils::explode( '-{', $txt );
 979                 $ret = array();
 980                 $first = true;
 981                 foreach ( $blocks as $block ) {
 982                         if ( $first ) {
 983                                 // Skip the part before the first -{
 984                                 $first = false;
 985                                 continue;
 986                         }
 987                         $mappings = explode( '}-', $block, 2 );
 988                         $stripped = str_replace( array( "'", '"', '*', '#' ), '', $mappings[0] );
 989                         $table = StringUtils::explode( ';', $stripped );
 990                         foreach ( $table as $t ) {
 991                                 $m = explode( '=>', $t, 3 );
 992                                 if ( count( $m ) != 2 ) {
 993                                         continue;
 994                                 }
 995                                 // trim any trailling comments starting with '//'
 996                                 $tt = explode( '//', $m[1], 2 );
 997                                 $ret[trim( $m[0] )] = trim( $tt[0] );
 998                         }
 999                 }
1000
1001                 // recursively parse the subpages
1002                 if ( $recursive ) {
1003                         foreach ( $sublinks as $link ) {
1004                                 $s = $this->parseCachedTable( $code, $link, $recursive );
1005                                 $ret = array_merge( $ret, $s );
1006                         }
1007                 }
1008
1009                 if ( $this->mUcfirst ) {
1010                         foreach ( $ret as $k => $v ) {
1011                                 $ret[$this->mLangObj->ucfirst( $k )] = $this->mLangObj->ucfirst( $v );
1012                         }
1013                 }
1014                 return $ret;
1015         }
1016
1017         /**
1018          * Enclose a string with the "no conversion" tag. This is used by
1019          * various functions in the Parser.
1020          *
1021          * @param string $text Text to be tagged for no conversion
1022          * @param bool $noParse Unused
1023          * @return string The tagged text
1024          */
1025         public function markNoConversion( $text, $noParse = false ) {
1026                 # don't mark if already marked
1027                 if ( strpos( $text, '-{' ) || strpos( $text, '}-' ) ) {
1028                         return $text;
1029                 }
1030
1031                 $ret = "-{R|$text}-";
1032                 return $ret;
1033         }
1034
1035         /**
1036          * Convert the sorting key for category links. This should make different
1037          * keys that are variants of each other map to the same key.
1038          *
1039          * @param $key string
1040          *
1041          * @return string
1042          */
1043         function convertCategoryKey( $key ) {
1044                 return $key;
1045         }
1046
1047         /**
1048          * Hook to refresh the cache of conversion tables when
1049          * MediaWiki:Conversiontable* is updated.
1050          * @private
1051          *
1052          * @param WikiPage $page
1053          * @param User $user User object for the current user
1054          * @param Content $content new page content
1055          * @param string $summary edit summary of the edit
1056          * @param bool $isMinor was the edit marked as minor?
1057          * @param bool $isWatch did the user watch this page or not?
1058          * @param string|int $section
1059          * @param int $flags Bitfield
1060          * @param Revision $revision new Revision object or null
1061          * @return bool true
1062          */
1063         function OnPageContentSaveComplete( $page, $user, $content, $summary, $isMinor,
1064                         $isWatch, $section, $flags, $revision ) {
1065                 $titleobj = $page->getTitle();
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                 return true;
1077         }
1078
1079         /**
1080          * Armour rendered math against conversion.
1081          * Escape special chars in parsed math text. (in most cases are img elements)
1082          *
1083          * @param string $text Text to armour against conversion
1084          * @return string Armoured text where { and } have been converted to
1085          *   &#123; and &#125;
1086          * @deprecated since 1.22 is no longer used
1087          */
1088         public function armourMath( $text ) {
1089                 // convert '-{' and '}-' to '-&#123;' and '&#125;-' to prevent
1090                 // any unwanted markup appearing in the math image tag.
1091                 $text = strtr( $text, array( '-{' => '-&#123;', '}-' => '&#125;-' ) );
1092                 return $text;
1093         }
1094
1095         /**
1096          * Get the cached separator pattern for ConverterRule::parseRules()
1097          */
1098         function getVarSeparatorPattern() {
1099                 if ( is_null( $this->mVarSeparatorPattern ) ) {
1100                         // varsep_pattern for preg_split:
1101                         // text should be splited by ";" only if a valid variant
1102                         // name exist after the markup, for example:
1103                         //  -{zh-hans:<span style="font-size:120%;">xxx</span>;zh-hant:\
1104                         //      <span style="font-size:120%;">yyy</span>;}-
1105                         // we should split it as:
1106                         //  array(
1107                         //        [0] => 'zh-hans:<span style="font-size:120%;">xxx</span>'
1108                         //        [1] => 'zh-hant:<span style="font-size:120%;">yyy</span>'
1109                         //        [2] => ''
1110                         //       )
1111                         $pat = '/;\s*(?=';
1112                         foreach ( $this->mVariants as $variant ) {
1113                                 // zh-hans:xxx;zh-hant:yyy
1114                                 $pat .= $variant . '\s*:|';
1115                                 // xxx=>zh-hans:yyy; xxx=>zh-hant:zzz
1116                                 $pat .= '[^;]*?=>\s*' . $variant . '\s*:|';
1117                         }
1118                         $pat .= '\s*$)/';
1119                         $this->mVarSeparatorPattern = $pat;
1120                 }
1121                 return $this->mVarSeparatorPattern;
1122         }
1123 }