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  * @author fdcn <fdcn64@gmail.com>
  28  * @author shinjiman <shinjiman@gmail.com>
  29  * @author PhiLiP <philip.npc@gmail.com>
  30  */
  31 class LanguageConverter {
  32         /**
  33          * languages supporting variants
  34          * @since 1.20
  35          * @var array
  36          */
  37         static public $languagesWithVariants = array(
  38                 'gan',
  39                 'iu',
  40                 'kk',
  41                 'ku',
  42                 'shi',
  43                 'sr',
  44                 'tg',
  45                 'uz',
  46                 'zh',
  47         );
  48
  49         public $mMainLanguageCode;
  50         public $mVariants, $mVariantFallbacks, $mVariantNames;
  51         public $mTablesLoaded = false;
  52         public $mTables;
  53         // 'bidirectional' 'unidirectional' 'disable' for each variant
  54         public $mManualLevel;
  55
  56         /**
  57          * @var string Memcached key name
  58          */
  59         public $mCacheKey;
  60
  61         public $mLangObj;
  62         public $mFlags;
  63         public $mDescCodeSep = ':', $mDescVarSep = ';';
  64         public $mUcfirst = false;
  65         public $mConvRuleTitle = false;
  66         public $mURLVariant;
  67         public $mUserVariant;
  68         public $mHeaderVariant;
  69         public $mMaxDepth = 10;
  70         public $mVarSeparatorPattern;
  71
  72         const CACHE_VERSION_KEY = 'VERSION 7';
  73
  74         /**
  75          * Constructor
  76          *
  77          * @param Language $langobj
  78          * @param string $maincode The main language code of this language
  79          * @param array $variants The supported variants of this language
  80          * @param array $variantfallbacks The fallback language of each variant
  81          * @param array $flags Defining the custom strings that maps to the flags
  82          * @param array $manualLevel Limit for supported variants
  83          */
  84         public function __construct( $langobj, $maincode, $variants = array(),
  85                                                                 $variantfallbacks = array(), $flags = array(),
  86                                                                 $manualLevel = array() ) {
  87                 global $wgDisabledVariants;
  88                 $this->mLangObj = $langobj;
  89                 $this->mMainLanguageCode = $maincode;
  90                 $this->mVariants = array_diff( $variants, $wgDisabledVariants );
  91                 $this->mVariantFallbacks = $variantfallbacks;
  92                 $this->mVariantNames = Language::fetchLanguageNames();
  93                 $this->mCacheKey = wfMemcKey( 'conversiontables', $maincode );
  94                 $defaultflags = array(
  95                         // 'S' show converted text
  96                         // '+' add rules for alltext
  97                         // 'E' the gave flags is error
  98                         // these flags above are reserved for program
  99                         'A' => 'A',       // add rule for convert code (all text convert)
 100                         'T' => 'T',       // title convert
 101                         'R' => 'R',       // raw content
 102                         'D' => 'D',       // convert description (subclass implement)
 103                         '-' => '-',       // remove convert (not implement)
 104                         'H' => 'H',       // add rule for convert code (but no display in placed code)
 105                         'N' => 'N'        // current variant name
 106                 );
 107                 $this->mFlags = array_merge( $defaultflags, $flags );
 108                 foreach ( $this->mVariants as $v ) {
 109                         if ( array_key_exists( $v, $manualLevel ) ) {
 110                                 $this->mManualLevel[$v] = $manualLevel[$v];
 111                         } else {
 112                                 $this->mManualLevel[$v] = 'bidirectional';
 113                         }
 114                         $this->mFlags[$v] = $v;
 115                 }
 116         }
 117
 118         /**
 119          * Get all valid variants.
 120          * Call this instead of using $this->mVariants directly.
 121          *
 122          * @return array Contains all valid variants
 123          */
 124         public function getVariants() {
 125                 return $this->mVariants;
 126         }
 127
 128         /**
 129          * In case some variant is not defined in the markup, we need
 130          * to have some fallback. For example, in zh, normally people
 131          * will define zh-hans and zh-hant, but less so for zh-sg or zh-hk.
 132          * when zh-sg is preferred but not defined, we will pick zh-hans
 133          * in this case. Right now this is only used by zh.
 134          *
 135          * @param string $variant The language code of the variant
 136          * @return string|array The code of the fallback language or the
 137          *   main code if there is no fallback
 138          */
 139         public function getVariantFallbacks( $variant ) {
 140                 if ( isset( $this->mVariantFallbacks[$variant] ) ) {
 141                         return $this->mVariantFallbacks[$variant];
 142                 }
 143                 return $this->mMainLanguageCode;
 144         }
 145
 146         /**
 147          * Get the title produced by the conversion rule.
 148          * @return string The converted title text
 149          */
 150         public function getConvRuleTitle() {
 151                 return $this->mConvRuleTitle;
 152         }
 153
 154         /**
 155          * Get preferred language variant.
 156          * @return string The preferred language code
 157          */
 158         public function getPreferredVariant() {
 159                 global $wgDefaultLanguageVariant, $wgUser;
 160
 161                 $req = $this->getURLVariant();
 162
 163                 if ( $wgUser->isLoggedIn() && !$req ) {
 164                         $req = $this->getUserVariant();
 165                 } elseif ( !$req ) {
 166                         $req = $this->getHeaderVariant();
 167                 }
 168
 169                 if ( $wgDefaultLanguageVariant && !$req ) {
 170                         $req = $this->validateVariant( $wgDefaultLanguageVariant );
 171                 }
 172
 173                 // This function, unlike the other get*Variant functions, is
 174                 // not memoized (i.e. there return value is not cached) since
 175                 // new information might appear during processing after this
 176                 // is first called.
 177                 if ( $this->validateVariant( $req ) ) {
 178                         return $req;
 179                 }
 180                 return $this->mMainLanguageCode;
 181         }
 182
 183         /**
 184          * Get default variant.
 185          * This function would not be affected by user's settings
 186          * @return string The default variant code
 187          */
 188         public function getDefaultVariant() {
 189                 global $wgDefaultLanguageVariant;
 190
 191                 $req = $this->getURLVariant();
 192
 193                 if ( !$req ) {
 194                         $req = $this->getHeaderVariant();
 195                 }
 196
 197                 if ( $wgDefaultLanguageVariant && !$req ) {
 198                         $req = $this->validateVariant( $wgDefaultLanguageVariant );
 199                 }
 200
 201                 if ( $req ) {
 202                         return $req;
 203                 }
 204                 return $this->mMainLanguageCode;
 205         }
 206
 207         /**
 208          * Validate the variant
 209          * @param string $variant The variant to validate
 210          * @return mixed Returns the variant if it is valid, null otherwise
 211          */
 212         public function validateVariant( $variant = null ) {
 213                 if ( $variant !== null && in_array( $variant, $this->mVariants ) ) {
 214                         return $variant;
 215                 }
 216                 return null;
 217         }
 218
 219         /**
 220          * Get the variant specified in the URL
 221          *
 222          * @return mixed Variant if one found, false otherwise.
 223          */
 224         public function getURLVariant() {
 225                 global $wgRequest;
 226
 227                 if ( $this->mURLVariant ) {
 228                         return $this->mURLVariant;
 229                 }
 230
 231                 // see if the preference is set in the request
 232                 $ret = $wgRequest->getText( 'variant' );
 233
 234                 if ( !$ret ) {
 235                         $ret = $wgRequest->getVal( 'uselang' );
 236                 }
 237
 238                 $this->mURLVariant = $this->validateVariant( $ret );
 239                 return $this->mURLVariant;
 240         }
 241
 242         /**
 243          * Determine if the user has a variant set.
 244          *
 245          * @return mixed Variant if one found, false otherwise.
 246          */
 247         protected function getUserVariant() {
 248                 global $wgUser, $wgContLang;
 249
 250                 // memoizing this function wreaks havoc on parserTest.php
 251                 /*
 252                 if ( $this->mUserVariant ) {
 253                         return $this->mUserVariant;
 254                 }
 255                 */
 256
 257                 // Get language variant preference from logged in users
 258                 // Don't call this on stub objects because that causes infinite
 259                 // recursion during initialisation
 260                 if ( $wgUser->isLoggedIn() ) {
 261                         if ( $this->mMainLanguageCode == $wgContLang->getCode() ) {
 262                                 $ret = $wgUser->getOption( 'variant' );
 263                         } else {
 264                                 $ret = $wgUser->getOption( 'variant-' . $this->mMainLanguageCode );
 265                         }
 266                 } else {
 267                         // figure out user lang without constructing wgLang to avoid
 268                         // infinite recursion
 269                         $ret = $wgUser->getOption( 'language' );
 270                 }
 271
 272                 $this->mUserVariant = $this->validateVariant( $ret );
 273                 return $this->mUserVariant;
 274         }
 275
 276         /**
 277          * Determine the language variant from the Accept-Language header.
 278          *
 279          * @return mixed Variant if one found, false otherwise.
 280          */
 281         protected function getHeaderVariant() {
 282                 global $wgRequest;
 283
 284                 if ( $this->mHeaderVariant ) {
 285                         return $this->mHeaderVariant;
 286                 }
 287
 288                 // see if some supported language variant is set in the
 289                 // HTTP header.
 290                 $languages = array_keys( $wgRequest->getAcceptLang() );
 291                 if ( empty( $languages ) ) {
 292                         return null;
 293                 }
 294
 295                 $fallbackLanguages = array();
 296                 foreach ( $languages as $language ) {
 297                         $this->mHeaderVariant = $this->validateVariant( $language );
 298                         if ( $this->mHeaderVariant ) {
 299                                 break;
 300                         }
 301
 302                         // To see if there are fallbacks of current language.
 303                         // We record these fallback variants, and process
 304                         // them later.
 305                         $fallbacks = $this->getVariantFallbacks( $language );
 306                         if ( is_string( $fallbacks ) && $fallbacks !== $this->mMainLanguageCode ) {
 307                                 $fallbackLanguages[] = $fallbacks;
 308                         } elseif ( is_array( $fallbacks ) ) {
 309                                 $fallbackLanguages =
 310                                         array_merge( $fallbackLanguages, $fallbacks );
 311                         }
 312                 }
 313
 314                 if ( !$this->mHeaderVariant ) {
 315                         // process fallback languages now
 316                         $fallback_languages = array_unique( $fallbackLanguages );
 317                         foreach ( $fallback_languages as $language ) {
 318                                 $this->mHeaderVariant = $this->validateVariant( $language );
 319                                 if ( $this->mHeaderVariant ) {
 320                                         break;
 321                                 }
 322                         }
 323                 }
 324
 325                 return $this->mHeaderVariant;
 326         }
 327
 328         /**
 329          * Dictionary-based conversion.
 330          * This function would not parse the conversion rules.
 331          * If you want to parse rules, try to use convert() or
 332          * convertTo().
 333          *
 334          * @param string $text The text to be converted
 335          * @param bool|string $toVariant The target language code
 336          * @return string The converted text
 337          */
 338         public function autoConvert( $text, $toVariant = false ) {
 339
 340                 $this->loadTables();
 341
 342                 if ( !$toVariant ) {
 343                         $toVariant = $this->getPreferredVariant();
 344                         if ( !$toVariant ) {
 345                                 return $text;
 346                         }
 347                 }
 348
 349                 if ( $this->guessVariant( $text, $toVariant ) ) {
 350                         return $text;
 351                 }
 352
 353                 /* we convert everything except:
 354                    1. HTML markups (anything between < and >)
 355                    2. HTML entities
 356                    3. placeholders created by the parser
 357                 */
 358                 $marker = '|' . Parser::MARKER_PREFIX . '[\-a-zA-Z0-9]+';
 359
 360                 // this one is needed when the text is inside an HTML markup
 361                 $htmlfix = '|<[^>]+$|^[^<>]*>';
 362
 363                 // disable convert to variants between <code> tags
 364                 $codefix = '<code>.+?<\/code>|';
 365                 // disable conversion of <script> tags
 366                 $scriptfix = '<script.*?>.*?<\/script>|';
 367                 // disable conversion of <pre> tags
 368                 $prefix = '<pre.*?>.*?<\/pre>|';
 369
 370                 $reg = '/' . $codefix . $scriptfix . $prefix .
 371                         '<[^>]+>|&[a-zA-Z#][a-z0-9]+;' . $marker . $htmlfix . '/s';
 372                 $startPos = 0;
 373                 $sourceBlob = '';
 374                 $literalBlob = '';
 375
 376                 // Guard against delimiter nulls in the input
 377                 $text = str_replace( "\000", '', $text );
 378
 379                 $markupMatches = null;
 380                 $elementMatches = null;
 381                 while ( $startPos < strlen( $text ) ) {
 382                         if ( preg_match( $reg, $text, $markupMatches, PREG_OFFSET_CAPTURE, $startPos ) ) {
 383                                 $elementPos = $markupMatches[0][1];
 384                                 $element = $markupMatches[0][0];
 385                         } else {
 386                                 $elementPos = strlen( $text );
 387                                 $element = '';
 388                         }
 389
 390                         // Queue the part before the markup for translation in a batch
 391                         $sourceBlob .= substr( $text, $startPos, $elementPos - $startPos ) . "\000";
 392
 393                         // Advance to the next position
 394                         $startPos = $elementPos + strlen( $element );
 395
 396                         // Translate any alt or title attributes inside the matched element
 397                         if ( $element !== ''
 398                                 && preg_match( '/^(<[^>\s]*)\s([^>]*)(.*)$/', $element, $elementMatches )
 399                         ) {
 400                                 $attrs = Sanitizer::decodeTagAttributes( $elementMatches[2] );
 401                                 $changed = false;
 402                                 foreach ( array( 'title', 'alt' ) as $attrName ) {
 403                                         if ( !isset( $attrs[$attrName] ) ) {
 404                                                 continue;
 405                                         }
 406                                         $attr = $attrs[$attrName];
 407                                         // Don't convert URLs
 408                                         if ( !strpos( $attr, '://' ) ) {
 409                                                 $attr = $this->recursiveConvertTopLevel( $attr, $toVariant );
 410                                         }
 411
 412                                         // Remove HTML tags to avoid disrupting the layout
 413                                         $attr = preg_replace( '/<[^>]+>/', '', $attr );
 414                                         if ( $attr !== $attrs[$attrName] ) {
 415                                                 $attrs[$attrName] = $attr;
 416                                                 $changed = true;
 417                                         }
 418                                 }
 419                                 if ( $changed ) {
 420                                         $element = $elementMatches[1] . Html::expandAttributes( $attrs ) .
 421                                                 $elementMatches[3];
 422                                 }
 423                         }
 424                         $literalBlob .= $element . "\000";
 425                 }
 426
 427                 // Do the main translation batch
 428                 $translatedBlob = $this->translate( $sourceBlob, $toVariant );
 429
 430                 // Put the output back together
 431                 $translatedIter = StringUtils::explode( "\000", $translatedBlob );
 432                 $literalIter = StringUtils::explode( "\000", $literalBlob );
 433                 $output = '';
 434                 while ( $translatedIter->valid() && $literalIter->valid() ) {
 435                         $output .= $translatedIter->current();
 436                         $output .= $literalIter->current();
 437                         $translatedIter->next();
 438                         $literalIter->next();
 439                 }
 440
 441                 return $output;
 442         }
 443
 444         /**
 445          * Translate a string to a variant.
 446          * Doesn't parse rules or do any of that other stuff, for that use
 447          * convert() or convertTo().
 448          *
 449          * @param string $text Text to convert
 450          * @param string $variant Variant language code
 451          * @return string Translated text
 452          */
 453         public function translate( $text, $variant ) {
 454                 // If $text is empty or only includes spaces, do nothing
 455                 // Otherwise translate it
 456                 if ( trim( $text ) ) {
 457                         $this->loadTables();
 458                         $text = $this->mTables[$variant]->replace( $text );
 459                 }
 460                 return $text;
 461         }
 462
 463         /**
 464          * Call translate() to convert text to all valid variants.
 465          *
 466          * @param string $text The text to be converted
 467          * @return array Variant => converted text
 468          */
 469         public function autoConvertToAllVariants( $text ) {
 470                 $this->loadTables();
 471
 472                 $ret = array();
 473                 foreach ( $this->mVariants as $variant ) {
 474                         $ret[$variant] = $this->translate( $text, $variant );
 475                 }
 476
 477                 return $ret;
 478         }
 479
 480         /**
 481          * Apply manual conversion rules.
 482          *
 483          * @param ConverterRule $convRule
 484          */
 485         protected function applyManualConv( $convRule ) {
 486                 // Use syntax -{T|zh-cn:TitleCN; zh-tw:TitleTw}- to custom
 487                 // title conversion.
 488                 // Bug 24072: $mConvRuleTitle was overwritten by other manual
 489                 // rule(s) not for title, this breaks the title conversion.
 490                 $newConvRuleTitle = $convRule->getTitle();
 491                 if ( $newConvRuleTitle ) {
 492                         // So I add an empty check for getTitle()
 493                         $this->mConvRuleTitle = $newConvRuleTitle;
 494                 }
 495
 496                 // merge/remove manual conversion rules to/from global table
 497                 $convTable = $convRule->getConvTable();
 498                 $action = $convRule->getRulesAction();
 499                 foreach ( $convTable as $variant => $pair ) {
 500                         if ( !$this->validateVariant( $variant ) ) {
 501                                 continue;
 502                         }
 503
 504                         if ( $action == 'add' ) {
 505                                 foreach ( $pair as $from => $to ) {
 506                                         // to ensure that $from and $to not be left blank
 507                                         // so $this->translate() could always return a string
 508                                         if ( $from || $to ) {
 509                                                 // more efficient than array_merge(), about 2.5 times.
 510                                                 $this->mTables[$variant]->setPair( $from, $to );
 511                                         }
 512                                 }
 513                         } elseif ( $action == 'remove' ) {
 514                                 $this->mTables[$variant]->removeArray( $pair );
 515                         }
 516                 }
 517         }
 518
 519         /**
 520          * Auto convert a Title object to a readable string in the
 521          * preferred variant.
 522          *
 523          * @param Title $title A object of Title
 524          * @return string Converted title text
 525          */
 526         public function convertTitle( $title ) {
 527                 $variant = $this->getPreferredVariant();
 528                 $index = $title->getNamespace();
 529                 if ( $index !== NS_MAIN ) {
 530                         $text = $this->convertNamespace( $index, $variant ) . ':';
 531                 } else {
 532                         $text = '';
 533                 }
 534                 $text .= $this->translate( $title->getText(), $variant );
 535                 return $text;
 536         }
 537
 538         /**
 539          * Get the namespace display name in the preferred variant.
 540          *
 541          * @param int $index Namespace id
 542          * @param string|null $variant Variant code or null for preferred variant
 543          * @return string Namespace name for display
 544          */
 545         public function convertNamespace( $index, $variant = null ) {
 546                 if ( $variant === null ) {
 547                         $variant = $this->getPreferredVariant();
 548                 }
 549                 if ( $index === NS_MAIN ) {
 550                         return '';
 551                 } else {
 552                         // First check if a message gives a converted name in the target variant.
 553                         $nsConvMsg = wfMessage( 'conversion-ns' . $index )->inLanguage( $variant );
 554                         if ( $nsConvMsg->exists() ) {
 555                                 return $nsConvMsg->plain();
 556                         }
 557                         // Then check if a message gives a converted name in content language
 558                         // which needs extra translation to the target variant.
 559                         $nsConvMsg = wfMessage( 'conversion-ns' . $index )->inContentLanguage();
 560                         if ( $nsConvMsg->exists() ) {
 561                                 return $this->translate( $nsConvMsg->plain(), $variant );
 562                         }
 563                         // No message exists, retrieve it from the target variant's namespace names.
 564                         $langObj = $this->mLangObj->factory( $variant );
 565                         return $langObj->getFormattedNsText( $index );
 566                 }
 567         }
 568
 569         /**
 570          * Convert text to different variants of a language. The automatic
 571          * conversion is done in autoConvert(). Here we parse the text
 572          * marked with -{}-, which specifies special conversions of the
 573          * text that can not be accomplished in autoConvert().
 574          *
 575          * Syntax of the markup:
 576          * -{code1:text1;code2:text2;...}-  or
 577          * -{flags|code1:text1;code2:text2;...}-  or
 578          * -{text}- in which case no conversion should take place for text
 579          *
 580          * @param string $text Text to be converted
 581          * @return string Converted text
 582          */
 583         public function convert( $text ) {
 584                 $variant = $this->getPreferredVariant();
 585                 return $this->convertTo( $text, $variant );
 586         }
 587
 588         /**
 589          * Same as convert() except a extra parameter to custom variant.
 590          *
 591          * @param string $text Text to be converted
 592          * @param string $variant The target variant code
 593          * @return string Converted text
 594          */
 595         public function convertTo( $text, $variant ) {
 596                 global $wgDisableLangConversion;
 597                 if ( $wgDisableLangConversion ) {
 598                         return $text;
 599                 }
 600                 // Reset converter state for a new converter run.
 601                 $this->mConvRuleTitle = false;
 602                 return $this->recursiveConvertTopLevel( $text, $variant );
 603         }
 604
 605         /**
 606          * Recursively convert text on the outside. Allow to use nested
 607          * markups to custom rules.
 608          *
 609          * @param string $text Text to be converted
 610          * @param string $variant The target variant code
 611          * @param int $depth Depth of recursion
 612          * @return string Converted text
 613          */
 614         protected function recursiveConvertTopLevel( $text, $variant, $depth = 0 ) {
 615                 $startPos = 0;
 616                 $out = '';
 617                 $length = strlen( $text );
 618                 $shouldConvert = !$this->guessVariant( $text, $variant );
 619
 620                 while ( $startPos < $length ) {
 621                         $pos = strpos( $text, '-{', $startPos );
 622
 623                         if ( $pos === false ) {
 624                                 // No more markup, append final segment
 625                                 $fragment = substr( $text, $startPos );
 626                                 $out .= $shouldConvert ? $this->autoConvert( $fragment, $variant ) : $fragment;
 627                                 return $out;
 628                         }
 629
 630                         // Markup found
 631                         // Append initial segment
 632                         $fragment = substr( $text, $startPos, $pos - $startPos );
 633                         $out .= $shouldConvert ? $this->autoConvert( $fragment, $variant ) : $fragment;
 634
 635                         // Advance position
 636                         $startPos = $pos;
 637
 638                         // Do recursive conversion
 639                         $out .= $this->recursiveConvertRule( $text, $variant, $startPos, $depth + 1 );
 640                 }
 641
 642                 return $out;
 643         }
 644
 645         /**
 646          * Recursively convert text on the inside.
 647          *
 648          * @param string $text Text to be converted
 649          * @param string $variant The target variant code
 650          * @param int $startPos
 651          * @param int $depth Depth of recursion
 652          *
 653          * @throws MWException
 654          * @return string Converted text
 655          */
 656         protected function recursiveConvertRule( $text, $variant, &$startPos, $depth = 0 ) {
 657                 // Quick sanity check (no function calls)
 658                 if ( $text[$startPos] !== '-' || $text[$startPos + 1] !== '{' ) {
 659                         throw new MWException( __METHOD__ . ': invalid input string' );
 660                 }
 661
 662                 $startPos += 2;
 663                 $inner = '';
 664                 $warningDone = false;
 665                 $length = strlen( $text );
 666
 667                 while ( $startPos < $length ) {
 668                         $m = false;
 669                         preg_match( '/-\{|\}-/', $text, $m, PREG_OFFSET_CAPTURE, $startPos );
 670                         if ( !$m ) {
 671                                 // Unclosed rule
 672                                 break;
 673                         }
 674
 675                         $token = $m[0][0];
 676                         $pos = $m[0][1];
 677
 678                         // Markup found
 679                         // Append initial segment
 680                         $inner .= substr( $text, $startPos, $pos - $startPos );
 681
 682                         // Advance position
 683                         $startPos = $pos;
 684
 685                         switch ( $token ) {
 686                                 case '-{':
 687                                         // Check max depth
 688                                         if ( $depth >= $this->mMaxDepth ) {
 689                                                 $inner .= '-{';
 690                                                 if ( !$warningDone ) {
 691                                                         $inner .= '<span class="error">' .
 692                                                                 wfMessage( 'language-converter-depth-warning' )
 693                                                                         ->numParams( $this->mMaxDepth )->inContentLanguage()->text() .
 694                                                                 '</span>';
 695                                                         $warningDone = true;
 696                                                 }
 697                                                 $startPos += 2;
 698                                                 continue;
 699                                         }
 700                                         // Recursively parse another rule
 701                                         $inner .= $this->recursiveConvertRule( $text, $variant, $startPos, $depth + 1 );
 702                                         break;
 703                                 case '}-':
 704                                         // Apply the rule
 705                                         $startPos += 2;
 706                                         $rule = new ConverterRule( $inner, $this );
 707                                         $rule->parse( $variant );
 708                                         $this->applyManualConv( $rule );
 709                                         return $rule->getDisplay();
 710                                 default:
 711                                         throw new MWException( __METHOD__ . ': invalid regex match' );
 712                         }
 713                 }
 714
 715                 // Unclosed rule
 716                 if ( $startPos < $length ) {
 717                         $inner .= substr( $text, $startPos );
 718                 }
 719                 $startPos = $length;
 720                 return '-{' . $this->autoConvert( $inner, $variant );
 721         }
 722
 723         /**
 724          * If a language supports multiple variants, it is possible that
 725          * non-existing link in one variant actually exists in another variant.
 726          * This function tries to find it. See e.g. LanguageZh.php
 727          * The input parameters may be modified upon return
 728          *
 729          * @param string &$link The name of the link
 730          * @param Title &$nt The title object of the link
 731          * @param bool $ignoreOtherCond To disable other conditions when
 732          *   we need to transclude a template or update a category's link
 733          */
 734         public function findVariantLink( &$link, &$nt, $ignoreOtherCond = false ) {
 735                 # If the article has already existed, there is no need to
 736                 # check it again, otherwise it may cause a fault.
 737                 if ( is_object( $nt ) && $nt->exists() ) {
 738                         return;
 739                 }
 740
 741                 global $wgDisableLangConversion, $wgDisableTitleConversion, $wgRequest;
 742                 $isredir = $wgRequest->getText( 'redirect', 'yes' );
 743                 $action = $wgRequest->getText( 'action' );
 744                 if ( $action == 'edit' && $wgRequest->getBool( 'redlink' ) ) {
 745                         $action = 'view';
 746                 }
 747                 $linkconvert = $wgRequest->getText( 'linkconvert', 'yes' );
 748                 $disableLinkConversion = $wgDisableLangConversion
 749                         || $wgDisableTitleConversion;
 750                 $linkBatch = new LinkBatch();
 751
 752                 $ns = NS_MAIN;
 753
 754                 if ( $disableLinkConversion ||
 755                         ( !$ignoreOtherCond &&
 756                                 ( $isredir == 'no'
 757                                         || $action == 'edit'
 758                                         || $action == 'submit'
 759                                         || $linkconvert == 'no' ) ) ) {
 760                         return;
 761                 }
 762
 763                 if ( is_object( $nt ) ) {
 764                         $ns = $nt->getNamespace();
 765                 }
 766
 767                 $variants = $this->autoConvertToAllVariants( $link );
 768                 if ( !$variants ) { // give up
 769                         return;
 770                 }
 771
 772                 $titles = array();
 773
 774                 foreach ( $variants as $v ) {
 775                         if ( $v != $link ) {
 776                                 $varnt = Title::newFromText( $v, $ns );
 777                                 if ( !is_null( $varnt ) ) {
 778                                         $linkBatch->addObj( $varnt );
 779                                         $titles[] = $varnt;
 780                                 }
 781                         }
 782                 }
 783
 784                 // fetch all variants in single query
 785                 $linkBatch->execute();
 786
 787                 foreach ( $titles as $varnt ) {
 788                         if ( $varnt->getArticleID() > 0 ) {
 789                                 $nt = $varnt;
 790                                 $link = $varnt->getText();
 791                                 break;
 792                         }
 793                 }
 794         }
 795
 796         /**
 797          * Returns language specific hash options.
 798          *
 799          * @return string
 800          */
 801         public function getExtraHashOptions() {
 802                 $variant = $this->getPreferredVariant();
 803
 804                 return '!' . $variant;
 805         }
 806
 807         /**
 808          * Guess if a text is written in a variant. This should be implemented in subclasses.
 809          *
 810          * @param string $text The text to be checked
 811          * @param string $variant Language code of the variant to be checked for
 812          * @return bool True if $text appears to be written in $variant, false if not
 813          *
 814          * @author Nikola Smolenski <smolensk@eunet.rs>
 815          * @since 1.19
 816          */
 817         public function guessVariant( $text, $variant ) {
 818                 return false;
 819         }
 820
 821         /**
 822          * Load default conversion tables.
 823          * This method must be implemented in derived class.
 824          *
 825          * @private
 826          * @throws MWException
 827          */
 828         function loadDefaultTables() {
 829                 $name = get_class( $this );
 830
 831                 throw new MWException( "Must implement loadDefaultTables() method in class $name" );
 832         }
 833
 834         /**
 835          * Load conversion tables either from the cache or the disk.
 836          * @private
 837          * @param bool $fromCache Load from memcached? Defaults to true.
 838          */
 839         function loadTables( $fromCache = true ) {
 840                 global $wgLanguageConverterCacheType;
 841
 842                 if ( $this->mTablesLoaded ) {
 843                         return;
 844                 }
 845
 846                 $this->mTablesLoaded = true;
 847                 $this->mTables = false;
 848                 $cache = ObjectCache::getInstance( $wgLanguageConverterCacheType );
 849                 if ( $fromCache ) {
 850                         wfProfileIn( __METHOD__ . '-cache' );
 851                         $this->mTables = $cache->get( $this->mCacheKey );
 852                         wfProfileOut( __METHOD__ . '-cache' );
 853                 }
 854                 if ( !$this->mTables || !array_key_exists( self::CACHE_VERSION_KEY, $this->mTables ) ) {
 855                         wfProfileIn( __METHOD__ . '-recache' );
 856                         // not in cache, or we need a fresh reload.
 857                         // We will first load the default tables
 858                         // then update them using things in MediaWiki:Conversiontable/*
 859                         $this->loadDefaultTables();
 860                         foreach ( $this->mVariants as $var ) {
 861                                 $cached = $this->parseCachedTable( $var );
 862                                 $this->mTables[$var]->mergeArray( $cached );
 863                         }
 864
 865                         $this->postLoadTables();
 866                         $this->mTables[self::CACHE_VERSION_KEY] = true;
 867
 868                         $cache->set( $this->mCacheKey, $this->mTables, 43200 );
 869                         wfProfileOut( __METHOD__ . '-recache' );
 870                 }
 871         }
 872
 873         /**
 874          * Hook for post processing after conversion tables are loaded.
 875          */
 876         function postLoadTables() {
 877         }
 878
 879         /**
 880          * Reload the conversion tables.
 881          *
 882          * @private
 883          */
 884         function reloadTables() {
 885                 if ( $this->mTables ) {
 886                         unset( $this->mTables );
 887                 }
 888
 889                 $this->mTablesLoaded = false;
 890                 $this->loadTables( false );
 891         }
 892
 893         /**
 894          * Parse the conversion table stored in the cache.
 895          *
 896          * The tables should be in blocks of the following form:
 897          *              -{
 898          *                      word => word ;
 899          *                      word => word ;
 900          *                      ...
 901          *              }-
 902          *
 903          * To make the tables more manageable, subpages are allowed
 904          * and will be parsed recursively if $recursive == true.
 905          *
 906          * @param string $code Language code
 907          * @param string $subpage Subpage name
 908          * @param bool $recursive Parse subpages recursively? Defaults to true.
 909          *
 910          * @return array
 911          */
 912         function parseCachedTable( $code, $subpage = '', $recursive = true ) {
 913                 static $parsed = array();
 914
 915                 $key = 'Conversiontable/' . $code;
 916                 if ( $subpage ) {
 917                         $key .= '/' . $subpage;
 918                 }
 919                 if ( array_key_exists( $key, $parsed ) ) {
 920                         return array();
 921                 }
 922
 923                 $parsed[$key] = true;
 924
 925                 if ( $subpage === '' ) {
 926                         $txt = MessageCache::singleton()->getMsgFromNamespace( $key, $code );
 927                 } else {
 928                         $txt = false;
 929                         $title = Title::makeTitleSafe( NS_MEDIAWIKI, $key );
 930                         if ( $title && $title->exists() ) {
 931                                 $revision = Revision::newFromTitle( $title );
 932                                 if ( $revision ) {
 933                                         if ( $revision->getContentModel() == CONTENT_MODEL_WIKITEXT ) {
 934                                                 $txt = $revision->getContent( Revision::RAW )->getNativeData();
 935                                         }
 936
 937                                         // @todo in the future, use a specialized content model, perhaps based on json!
 938                                 }
 939                         }
 940                 }
 941
 942                 # Nothing to parse if there's no text
 943                 if ( $txt === false || $txt === null || $txt === '' ) {
 944                         return array();
 945                 }
 946
 947                 // get all subpage links of the form
 948                 // [[MediaWiki:Conversiontable/zh-xx/...|...]]
 949                 $linkhead = $this->mLangObj->getNsText( NS_MEDIAWIKI ) .
 950                         ':Conversiontable';
 951                 $subs = StringUtils::explode( '[[', $txt );
 952                 $sublinks = array();
 953                 foreach ( $subs as $sub ) {
 954                         $link = explode( ']]', $sub, 2 );
 955                         if ( count( $link ) != 2 ) {
 956                                 continue;
 957                         }
 958                         $b = explode( '|', $link[0], 2 );
 959                         $b = explode( '/', trim( $b[0] ), 3 );
 960                         if ( count( $b ) == 3 ) {
 961                                 $sublink = $b[2];
 962                         } else {
 963                                 $sublink = '';
 964                         }
 965
 966                         if ( $b[0] == $linkhead && $b[1] == $code ) {
 967                                 $sublinks[] = $sublink;
 968                         }
 969                 }
 970
 971                 // parse the mappings in this page
 972                 $blocks = StringUtils::explode( '-{', $txt );
 973                 $ret = array();
 974                 $first = true;
 975                 foreach ( $blocks as $block ) {
 976                         if ( $first ) {
 977                                 // Skip the part before the first -{
 978                                 $first = false;
 979                                 continue;
 980                         }
 981                         $mappings = explode( '}-', $block, 2 );
 982                         $stripped = str_replace( array( "'", '"', '*', '#' ), '', $mappings[0] );
 983                         $table = StringUtils::explode( ';', $stripped );
 984                         foreach ( $table as $t ) {
 985                                 $m = explode( '=>', $t, 3 );
 986                                 if ( count( $m ) != 2 ) {
 987                                         continue;
 988                                 }
 989                                 // trim any trailling comments starting with '//'
 990                                 $tt = explode( '//', $m[1], 2 );
 991                                 $ret[trim( $m[0] )] = trim( $tt[0] );
 992                         }
 993                 }
 994
 995                 // recursively parse the subpages
 996                 if ( $recursive ) {
 997                         foreach ( $sublinks as $link ) {
 998                                 $s = $this->parseCachedTable( $code, $link, $recursive );
 999                                 $ret = array_merge( $ret, $s );
1000                         }
1001                 }
1002
1003                 if ( $this->mUcfirst ) {
1004                         foreach ( $ret as $k => $v ) {
1005                                 $ret[$this->mLangObj->ucfirst( $k )] = $this->mLangObj->ucfirst( $v );
1006                         }
1007                 }
1008                 return $ret;
1009         }
1010
1011         /**
1012          * Enclose a string with the "no conversion" tag. This is used by
1013          * various functions in the Parser.
1014          *
1015          * @param string $text Text to be tagged for no conversion
1016          * @param bool $noParse Unused
1017          * @return string The tagged text
1018          */
1019         public function markNoConversion( $text, $noParse = false ) {
1020                 # don't mark if already marked
1021                 if ( strpos( $text, '-{' ) || strpos( $text, '}-' ) ) {
1022                         return $text;
1023                 }
1024
1025                 $ret = "-{R|$text}-";
1026                 return $ret;
1027         }
1028
1029         /**
1030          * Convert the sorting key for category links. This should make different
1031          * keys that are variants of each other map to the same key.
1032          *
1033          * @param string $key
1034          *
1035          * @return string
1036          */
1037         function convertCategoryKey( $key ) {
1038                 return $key;
1039         }
1040
1041         /**
1042          * Hook to refresh the cache of conversion tables when
1043          * MediaWiki:Conversiontable* is updated.
1044          * @private
1045          *
1046          * @param WikiPage $page
1047          * @param User $user User object for the current user
1048          * @param Content $content New page content
1049          * @param string $summary Edit summary of the edit
1050          * @param bool $isMinor Was the edit marked as minor?
1051          * @param null $isWatch Unused.
1052          * @param null $section Unused.
1053          * @param int $flags Bitfield
1054          * @param Revision|null $revision New Revision object or null
1055          * @return bool True
1056          */
1057         function OnPageContentSaveComplete( $page, $user, $content, $summary, $isMinor,
1058                         $isWatch, $section, $flags, $revision ) {
1059                 $titleobj = $page->getTitle();
1060                 if ( $titleobj->getNamespace() == NS_MEDIAWIKI ) {
1061                         $title = $titleobj->getDBkey();
1062                         $t = explode( '/', $title, 3 );
1063                         $c = count( $t );
1064                         if ( $c > 1 && $t[0] == 'Conversiontable' ) {
1065                                 if ( $this->validateVariant( $t[1] ) ) {
1066                                         $this->reloadTables();
1067                                 }
1068                         }
1069                 }
1070                 return true;
1071         }
1072
1073         /**
1074          * Armour rendered math against conversion.
1075          * Escape special chars in parsed math text. (in most cases are img elements)
1076          *
1077          * @param string $text Text to armour against conversion
1078          * @return string Armoured text where { and } have been converted to
1079          *   &#123; and &#125;
1080          * @deprecated since 1.22 is no longer used
1081          */
1082         public function armourMath( $text ) {
1083                 // convert '-{' and '}-' to '-&#123;' and '&#125;-' to prevent
1084                 // any unwanted markup appearing in the math image tag.
1085                 $text = strtr( $text, array( '-{' => '-&#123;', '}-' => '&#125;-' ) );
1086                 return $text;
1087         }
1088
1089         /**
1090          * Get the cached separator pattern for ConverterRule::parseRules()
1091          * @return string
1092          */
1093         function getVarSeparatorPattern() {
1094                 if ( is_null( $this->mVarSeparatorPattern ) ) {
1095                         // varsep_pattern for preg_split:
1096                         // text should be splited by ";" only if a valid variant
1097                         // name exist after the markup, for example:
1098                         //  -{zh-hans:<span style="font-size:120%;">xxx</span>;zh-hant:\
1099                         //      <span style="font-size:120%;">yyy</span>;}-
1100                         // we should split it as:
1101                         //  array(
1102                         //        [0] => 'zh-hans:<span style="font-size:120%;">xxx</span>'
1103                         //        [1] => 'zh-hant:<span style="font-size:120%;">yyy</span>'
1104                         //        [2] => ''
1105                         //       )
1106                         $pat = '/;\s*(?=';
1107                         foreach ( $this->mVariants as $variant ) {
1108                                 // zh-hans:xxx;zh-hant:yyy
1109                                 $pat .= $variant . '\s*:|';
1110                                 // xxx=>zh-hans:yyy; xxx=>zh-hant:zzz
1111                                 $pat .= '[^;]*?=>\s*' . $variant . '\s*:|';
1112                         }
1113                         $pat .= '\s*$)/';
1114                         $this->mVarSeparatorPattern = $pat;
1115                 }
1116                 return $this->mVarSeparatorPattern;
1117         }
1118 }