languages/LanguageConverter.php

   1 <?php
   2
   3 /**
   4  * Contains the LanguageConverter class and ConverterRule class
   5  * @ingroup Language
   6  *
   7  * @license http://www.gnu.org/copyleft/gpl.html GNU General Public License
   8  * @file
   9  */
  10
  11 /**
  12  * base class for language convert
  13  * @ingroup Language
  14  *
  15  * @author Zhengzhu Feng <zhengzhu@gmail.com>
  16  * @maintainers fdcn <fdcn64@gmail.com>, shinjiman <shinjiman@gmail.com>, PhiLiP <philip.npc@gmail.com>
  17  */
  18 class LanguageConverter {
  19         var $mPreferredVariant='';
  20         var $mMainLanguageCode;
  21         var $mVariants, $mVariantFallbacks, $mVariantNames;
  22         var $mTablesLoaded = false;
  23         var $mTables;
  24         var $mNamespaceTables;
  25         var $mTitleDisplay='';
  26         var $mDoTitleConvert=true, $mDoContentConvert=true;
  27         var $mManualLevel; // 'bidirectional' 'unidirectional' 'disable' for each variants
  28         var $mTitleFromFlag = false;
  29         var $mCacheKey;
  30         var $mLangObj;
  31         var $mMarkup;
  32         var $mFlags;
  33         var $mDescCodeSep = ':',$mDescVarSep = ';';
  34         var $mUcfirst = false;
  35
  36         const CACHE_VERSION_KEY = 'VERSION 6';
  37
  38         /**
  39          * Constructor
  40          *
  41          * @param string $maincode the main language code of this language
  42          * @param array $variants the supported variants of this language
  43          * @param array $variantfallback the fallback language of each variant
  44          * @param array $markup array defining the markup used for manual conversion
  45          * @param array $flags array defining the custom strings that maps to the flags
  46          * @param array $manualLevel limit for supported variants
  47          * @public
  48          */
  49         function __construct( $langobj, $maincode,
  50                                                                 $variants=array(),
  51                                                                 $variantfallbacks=array(),
  52                                                                 $markup=array(),
  53                                                                 $flags = array(),
  54                                                                 $manualLevel = array() ) {
  55                 $this->mLangObj = $langobj;
  56                 $this->mMainLanguageCode = $maincode;
  57
  58                 global $wgDisabledVariants;
  59                 $this->mVariants = array();
  60                 foreach( $variants as $variant ) {
  61                         if( !in_array( $variant, $wgDisabledVariants ) )
  62                                 $this->mVariants[] = $variant;
  63                 }
  64                 $this->mVariantFallbacks = $variantfallbacks;
  65                 global $wgLanguageNames;
  66                 $this->mVariantNames = $wgLanguageNames;
  67                 $this->mCacheKey = wfMemcKey( 'conversiontables', $maincode );
  68                 $m = array(
  69                         'begin'=>'-{',
  70                         'flagsep'=>'|',
  71                         'unidsep'=>'=>', //for unidirectional conversion
  72                         'codesep'=>':',
  73                         'varsep'=>';',
  74                         'end'=>'}-'
  75                 );
  76                 $this->mMarkup = array_merge($m, $markup);
  77                 $f = array(
  78                         // 'S' show converted text
  79                         // '+' add rules for alltext
  80                         // 'E' the gave flags is error
  81                         // these flags above are reserved for program
  82                         'A'=>'A',       // add rule for convert code (all text convert)
  83                         'T'=>'T',       // title convert
  84                         'R'=>'R',       // raw content
  85                         'D'=>'D',       // convert description (subclass implement)
  86                         '-'=>'-',       // remove convert (not implement)
  87                         'H'=>'H',       // add rule for convert code (but no display in placed code )
  88                         'N'=>'N'        // current variant name
  89                 );
  90                 $this->mFlags = array_merge($f, $flags);
  91                 foreach( $this->mVariants as $v) {
  92                         $this->mManualLevel[$v]=array_key_exists($v,$manualLevel)
  93                                                                 ?$manualLevel[$v]
  94                                                                 :'bidirectional';
  95                         $this->mNamespaceTables[$v] = array();
  96                         $this->mFlags[$v] = $v;
  97                 }
  98         }
  99
 100         /**
 101          * @public
 102          */
 103         function getVariants() {
 104                 return $this->mVariants;
 105         }
 106
 107         /**
 108          * in case some variant is not defined in the markup, we need
 109          * to have some fallback. for example, in zh, normally people
 110          * will define zh-hans and zh-hant, but less so for zh-sg or zh-hk.
 111          * when zh-sg is preferred but not defined, we will pick zh-hans
 112          * in this case. right now this is only used by zh.
 113          *
 114          * @param string $v the language code of the variant
 115          * @return string array the code of the fallback language or false if there is no fallback
 116          * @public
 117          */
 118         function getVariantFallbacks($v) {
 119                 if( isset( $this->mVariantFallbacks[$v] ) ) {
 120                         return $this->mVariantFallbacks[$v];
 121                 }
 122                 return $this->mMainLanguageCode;
 123         }
 124
 125         /**
 126          * get preferred language variants.
 127          * @param boolean $fromUser Get it from $wgUser's preferences
 128          * @return string the preferred language code
 129          * @public
 130          */
 131         function getPreferredVariant( $fromUser = true ) {
 132                 global $wgUser, $wgRequest, $wgVariantArticlePath, $wgDefaultLanguageVariant;
 133
 134                 if($this->mPreferredVariant)
 135                         return $this->mPreferredVariant;
 136
 137                 // figure out user lang without constructing wgLang to avoid infinite recursion
 138                 if( $fromUser )
 139                         $defaultUserLang = $wgUser->getOption( 'language' );
 140                 else
 141                         $defaultUserLang = $this->mMainLanguageCode;
 142                 $userLang = $wgRequest->getVal( 'uselang', $defaultUserLang );
 143                 // see if interface language is same as content, if not, prevent conversion
 144                 if( ! in_array( $userLang, $this->mVariants ) ){
 145                         $this->mPreferredVariant = $this->mMainLanguageCode; // no conversion
 146                         return $this->mPreferredVariant;
 147                 }
 148
 149                 // see if the preference is set in the request
 150                 $req = $wgRequest->getText( 'variant' );
 151                 if( in_array( $req, $this->mVariants ) ) {
 152                         $this->mPreferredVariant = $req;
 153                         return $req;
 154                 }
 155
 156                 // check the syntax /code/ArticleTitle
 157                 if($wgVariantArticlePath!=false && isset($_SERVER['SCRIPT_NAME'])){
 158                         // Note: SCRIPT_NAME probably won't hold the correct value if PHP is run as CGI
 159                         // (it will hold path to php.cgi binary), and might not exist on some very old PHP installations
 160                         $scriptBase = basename( $_SERVER['SCRIPT_NAME'] );
 161                         if(in_array($scriptBase,$this->mVariants)){
 162                                 $this->mPreferredVariant = $scriptBase;
 163                                 return $this->mPreferredVariant;
 164                         }
 165                 }
 166
 167                 // get language variant preference from logged in users
 168                 // Don't call this on stub objects because that causes infinite
 169                 // recursion during initialisation
 170                 if( $fromUser && $wgUser->isLoggedIn() )  {
 171                         $this->mPreferredVariant = $wgUser->getOption('variant');
 172                         return $this->mPreferredVariant;
 173                 }
 174
 175                 // see if default variant is globaly set
 176                 if($wgDefaultLanguageVariant != false  &&  in_array( $wgDefaultLanguageVariant, $this->mVariants )){
 177                         $this->mPreferredVariant = $wgDefaultLanguageVariant;
 178                         return $this->mPreferredVariant;
 179                 }
 180
 181                 if( !$this->mPreferredVariant ) {
 182                         // see if some supported language variant is set in the
 183                         // http header, but we don't set the mPreferredVariant
 184                         // variable in case this is called before the user's
 185                         // preference is loaded
 186                         if( array_key_exists( 'HTTP_ACCEPT_LANGUAGE', $_SERVER ) ) {
 187                                 $acceptLanguage = strtolower( $_SERVER['HTTP_ACCEPT_LANGUAGE'] );
 188
 189                                 // explode by comma
 190                                 $result = explode(',', $acceptLanguage);
 191
 192                                 $languages = array();
 193
 194                                 foreach( $result as $elem ) {
 195                                         // if $elem likes 'zh-cn;q=0.9'
 196                                         if(($posi = strpos( $elem, ';' )) !== false ) {
 197                                                 // get the real language code likes 'zh-cn'
 198                                                 $languages[] = substr( $elem, 0, $posi );
 199                                         }
 200                                         else {
 201                                                 $languages[] = $elem;
 202                                         }
 203                                 }
 204
 205                                 $fallback_languages = array();
 206                                 foreach( $languages as $language ) {
 207                                         // strip whitespace
 208                                         $language = trim( $language );
 209                                         if( in_array( $language, $this->mVariants ) ) {
 210                                                 return $language;
 211                                         }
 212                                         else {
 213                                                 // To see if there are fallbacks of current language.
 214                                                 // We record these fallback variants, and process
 215                                                 // them later.
 216                                                 $fallbacks = $this->getVariantFallbacks( $language );
 217                                                 if( is_string( $fallbacks ) )
 218                                                         $fallback_languages[] = $fallbacks;
 219                                                 elseif( is_array( $fallbacks ) )
 220                                                         $fallback_languages = array_merge( $fallback_languages, $fallbacks );
 221                                         }
 222                                 }
 223
 224                                 // process fallback languages now
 225                                 $fallback_languages = array_unique( $fallback_languages );
 226                                 foreach( $fallback_languages as $language ) {
 227                                         if( in_array( $language, $this->mVariants ) ) {
 228                                                 return $language;
 229                                         }
 230                                 }
 231                         }
 232                 }
 233
 234                 return $this->mMainLanguageCode;
 235         }
 236
 237         /**
 238          * caption convert, base on preg_replace_callback
 239          *
 240          * to convert text in "title" or "alt", like '<img alt="text" ... '
 241          * or '<span title="text" ... '
 242          *
 243          * @return string like ' alt="yyyy"' or ' title="yyyy"'
 244          * @private
 245          */
 246         function captionConvert( $matches ) {
 247                 $toVariant = $this->getPreferredVariant();
 248                 $title = $matches[1];
 249                 $text  = $matches[2];
 250                 // we convert captions except URL
 251                 if( !strpos( $text, '://' ) )
 252                         $text = $this->translate($text, $toVariant);
 253                 return " $title=\"$text\"";
 254         }
 255
 256         /**
 257          * dictionary-based conversion
 258          *
 259          * @param string $text the text to be converted
 260          * @param string $toVariant the target language code
 261          * @return string the converted text
 262          * @private
 263          */
 264         function autoConvert($text, $toVariant=false) {
 265                 $fname="LanguageConverter::autoConvert";
 266
 267                 wfProfileIn( $fname );
 268
 269                 if(!$this->mTablesLoaded)
 270                         $this->loadTables();
 271
 272                 if(!$toVariant)
 273                         $toVariant = $this->getPreferredVariant();
 274                 if(!in_array($toVariant, $this->mVariants))
 275                         return $text;
 276
 277                 /* we convert everything except:
 278                    1. html markups (anything between < and >)
 279                    2. html entities
 280                    3. place holders created by the parser
 281                 */
 282                 global $wgParser;
 283                 if (isset($wgParser) && $wgParser->UniqPrefix()!=''){
 284                         $marker = '|' . $wgParser->UniqPrefix() . '[\-a-zA-Z0-9]+';
 285                 } else
 286                         $marker = "";
 287
 288                 // this one is needed when the text is inside an html markup
 289                 $htmlfix = '|<[^>]+$|^[^<>]*>';
 290
 291                 // disable convert to variants between <code></code> tags
 292                 $codefix = '<code>.+?<\/code>|';
 293                 // disable convertsion of <script type="text/javascript"> ... </script>
 294                 $scriptfix = '<script.*?>.*?<\/script>|';
 295                 // disable conversion of <pre xxxx> ... </pre>
 296                 $prefix = '<pre.*?>.*?<\/pre>|';
 297
 298                 $reg = '/'.$codefix . $scriptfix . $prefix . '<[^>]+>|&[a-zA-Z#][a-z0-9]+;' . $marker . $htmlfix . '/s';
 299
 300                 $matches = preg_split($reg, $text, -1, PREG_SPLIT_OFFSET_CAPTURE);
 301
 302                 $m = array_shift($matches);
 303
 304                 $ret = $this->translate($m[0], $toVariant);
 305                 $mstart = $m[1]+strlen($m[0]);
 306
 307                 // enable convertsion of '<img alt="xxxx" ... ' or '<span title="xxxx" ... '
 308                 $captionpattern  = '/\s(title|alt)\s*=\s*"([\s\S]*?)"/';
 309                 foreach($matches as $m) {
 310                         $mark = substr($text, $mstart, $m[1]-$mstart);
 311                         $mark = preg_replace_callback($captionpattern, array(&$this, 'captionConvert'), $mark);
 312                         $ret .= $mark;
 313                         $ret .= $this->translate($m[0], $toVariant);
 314                         $mstart = $m[1] + strlen($m[0]);
 315                 }
 316                 wfProfileOut( $fname );
 317                 return $ret;
 318         }
 319
 320         /**
 321          * Translate a string to a variant
 322          * Doesn't process markup or do any of that other stuff, for that use convert()
 323          *
 324          * @param string $text Text to convert
 325          * @param string $variant Variant language code
 326          * @return string Translated text
 327          * @private
 328          */
 329         function translate( $text, $variant ) {
 330                 wfProfileIn( __METHOD__ );
 331                 if( !$this->mTablesLoaded )
 332                         $this->loadTables();
 333                 $text = $this->mTables[$variant]->replace( $text );
 334                 wfProfileOut( __METHOD__ );
 335                 return $text;
 336         }
 337
 338         /**
 339          * convert text to all supported variants
 340          *
 341          * @param string $text the text to be converted
 342          * @return array of string
 343          * @public
 344          */
 345         function autoConvertToAllVariants($text) {
 346                 $fname="LanguageConverter::autoConvertToAllVariants";
 347                 wfProfileIn( $fname );
 348                 if( !$this->mTablesLoaded )
 349                         $this->loadTables();
 350
 351                 $ret = array();
 352                 foreach($this->mVariants as $variant) {
 353                         $ret[$variant] = $this->translate($text, $variant);
 354                 }
 355
 356                 wfProfileOut( $fname );
 357                 return $ret;
 358         }
 359
 360         /**
 361          * convert link text to all supported variants
 362          *
 363          * @param string $text the text to be converted
 364          * @return array of string
 365          * @public
 366          */
 367         function convertLinkToAllVariants($text) {
 368                 if( !$this->mTablesLoaded )
 369                         $this->loadTables();
 370
 371                 $ret = array();
 372                 $tarray = explode($this->mMarkup['begin'], $text);
 373                 $tfirst = array_shift($tarray);
 374
 375                 foreach($this->mVariants as $variant)
 376                         $ret[$variant] = $this->translate($tfirst,$variant);
 377
 378                 foreach($tarray as $txt) {
 379                         $marked = explode($this->mMarkup['end'], $txt, 2);
 380
 381                         foreach($this->mVariants as $variant){
 382                                 $ret[$variant] .= $this->mMarkup['begin'].$marked[0].$this->mMarkup['end'];
 383                                 if(array_key_exists(1, $marked))
 384                                         $ret[$variant] .= $this->translate($marked[1],$variant);
 385                         }
 386
 387                 }
 388
 389                 return $ret;
 390         }
 391
 392         /**
 393          * prepare manual conversion table
 394          * @private
 395          */
 396         function applyManualConv( $convRule ){
 397                 // use syntax -{T|zh:TitleZh;zh-tw:TitleTw}- for custom conversion in title
 398                 $title = $convRule->getTitle();
 399                 if( $title ){
 400                         $this->mTitleFromFlag = true;
 401                         $this->mTitleDisplay =  $title;
 402                 }
 403
 404                 //apply manual conversion table to global table
 405                 $convTable = $convRule->getConvTable();
 406                 $action = $convRule->getRulesAction();
 407                 foreach( $convTable as $variant => $pair ) {
 408                         if( !in_array( $variant, $this->mVariants ) )continue;
 409                         if( $action=="add" ) {
 410                                 foreach( $pair as $from => $to ) {
 411                                         // to ensure that $from and $to not be left blank
 412                                         // so $this->translate() could always return a string
 413                                         if ( $from || $to )
 414                                                 // more efficient than array_merge(), about 2.5 times.
 415                                                 $this->mTables[$variant]->setPair( $from, $to );
 416                                 }
 417                         }
 418                         elseif ( $action == "remove" ) {
 419                                 $this->mTables[$variant]->removeArray( $pair );
 420                         }
 421                 }
 422         }
 423
 424         /**
 425          * Convert text using a parser object for context
 426          * @public
 427          */
 428         function parserConvert( $text, &$parser ) {
 429                 global $wgDisableLangConversion;
 430                 /* don't do anything if this is the conversion table */
 431                 if ( $parser->getTitle()->getNamespace() == NS_MEDIAWIKI &&
 432                                  strpos($parser->mTitle->getText(), "Conversiontable") !== false )
 433                 {
 434                         return $text;
 435                 }
 436
 437                 if ( $wgDisableLangConversion )
 438                         return $text;
 439
 440                 $text = $this->convert( $text );
 441
 442                 if ( $this->mTitleFromFlag )
 443                         $parser->mOutput->setTitleText( $this->mTitleDisplay );
 444                 return $text;
 445         }
 446
 447         /**
 448          * convert namespace
 449          * @param string $title the title included namespace
 450          * @return array of string
 451          * @private
 452          */
 453         function convertNamespace( $title, $variant ) {
 454                 $splittitle = explode( ':', $title );
 455                 if (count($splittitle) < 2)
 456                         return $title;
 457                 if ( isset( $this->mNamespaceTables[$variant][$splittitle[0]] ) )
 458                         $splittitle[0] = $this->mNamespaceTables[$variant][$splittitle[0]];
 459                 $ret = implode(':', $splittitle );
 460                 return $ret;
 461         }
 462
 463         /**
 464          * convert title
 465          * @private
 466          */
 467         function convertTitle( $text, $variant ){
 468                 global $wgDisableTitleConversion, $wgUser;
 469
 470                 // check for global param and __NOTC__ tag
 471                 if( $wgDisableTitleConversion || !$this->mDoTitleConvert || $wgUser->getOption('noconvertlink') == 1 ) {
 472                         $this->mTitleDisplay = $text;
 473                         return $text;
 474                 }
 475
 476                 // use the title from the T flag if any
 477                 if( $this->mTitleFromFlag ){
 478                         $this->mTitleFromFlag = false;
 479                         return $this->mTitleDisplay;
 480                 }
 481
 482                 global $wgRequest;
 483                 $isredir = $wgRequest->getText( 'redirect', 'yes' );
 484                 $action = $wgRequest->getText( 'action' );
 485                 $linkconvert = $wgRequest->getText( 'linkconvert', 'yes' );
 486                 if ( $isredir == 'no' || $action == 'edit' || $action == 'submit' || $linkconvert == 'no' ) {
 487                         return $text;
 488                 } else {
 489                         $text = $this->convertNamespace( $text, $variant );
 490                         $this->mTitleDisplay = $this->convert( $text );
 491                         return $this->mTitleDisplay;
 492                 }
 493         }
 494
 495         /**
 496          * convert text to different variants of a language. the automatic
 497          * conversion is done in autoConvert(). here we parse the text
 498          * marked with -{}-, which specifies special conversions of the
 499          * text that can not be accomplished in autoConvert()
 500          *
 501          * syntax of the markup:
 502          * -{code1:text1;code2:text2;...}-  or
 503          * -{flags|code1:text1;code2:text2;...}-  or
 504          * -{text}- in which case no conversion should take place for text
 505          *
 506          * @param string $text text to be converted
 507          * @param bool $isTitle whether this conversion is for the article title
 508          * @return string converted text
 509          * @public
 510          */
 511         function convert( $text, $isTitle = false ) {
 512
 513                 $mw =& MagicWord::get( 'notitleconvert' );
 514                 if( $mw->matchAndRemove( $text ) )
 515                         $this->mDoTitleConvert = false;
 516                 $mw =& MagicWord::get( 'nocontentconvert' );
 517                 if( $mw->matchAndRemove( $text ) ) {
 518                         $this->mDoContentConvert = false;
 519                 }
 520
 521                 // no conversion if redirecting
 522                 $mw =& MagicWord::get( 'redirect' );
 523                 if( $mw->matchStart( $text ) )
 524                         return $text;
 525
 526                 $plang = $this->getPreferredVariant();
 527
 528                 // for title convertion
 529                 if ( $isTitle ) return $this->convertTitle( $text, $plang );
 530
 531                 $tarray = StringUtils::explode( $this->mMarkup['end'], $text );
 532                 $text = '';
 533
 534                 foreach ( $tarray as $txt ) {
 535
 536                         $marked = explode( $this->mMarkup['begin'], $txt, 2 );
 537
 538                         if( $this->mDoContentConvert )
 539                                 // Bug 19620: should convert a string immediately after a new rule added.
 540                                 $text .= $this->autoConvert( $marked[0], $plang );
 541                         else
 542                                 $text .= $marked[0];
 543
 544                         if ( array_key_exists( 1, $marked ) ) {
 545                                 $crule = new ConverterRule($marked[1], $this);
 546                                 $crule->parse( $plang );
 547                                 $text .= $crule->getDisplay();
 548                                 $this->applyManualConv( $crule );
 549                         }
 550                         else
 551                                 $text .= $this->mMarkup['end'];
 552
 553                 }
 554
 555                 // Remove the last delimiter (wasn't real)
 556                 $text = substr( $text, 0, -strlen( $this->mMarkup['end'] ) );
 557                 return $text;
 558         }
 559
 560         /**
 561          * if a language supports multiple variants, it is
 562          * possible that non-existing link in one variant
 563          * actually exists in another variant. this function
 564          * tries to find it. See e.g. LanguageZh.php
 565          *
 566          * @param string $link the name of the link
 567          * @param mixed $nt the title object of the link
 568          * @param boolean $ignoreOtherCond: to disable other conditions when
 569          *      we need to transclude a template or update a category's link
 570          * @return null the input parameters may be modified upon return
 571          * @public
 572          */
 573         function findVariantLink( &$link, &$nt, $ignoreOtherCond = false ) {
 574                 # If the article has already existed, there is no need to
 575                 # check it again, otherwise it may cause a fault.
 576                 if ( is_object( $nt ) && $nt->exists() )
 577                         return;
 578
 579                 global $wgDisableLangConversion, $wgDisableTitleConversion, $wgRequest, $wgUser;
 580                 $isredir = $wgRequest->getText( 'redirect', 'yes' );
 581                 $action = $wgRequest->getText( 'action' );
 582                 $linkconvert = $wgRequest->getText( 'linkconvert', 'yes' );
 583                 $disableLinkConversion = $wgDisableLangConversion || $wgDisableTitleConversion;
 584                 $linkBatch = new LinkBatch();
 585
 586                 $ns=NS_MAIN;
 587
 588                 if ( $disableLinkConversion || ( !$ignoreOtherCond && ( $isredir == 'no' || $action == 'edit'
 589                         || $action == 'submit' || $linkconvert == 'no' || $wgUser->getOption('noconvertlink') == 1 ) ) )
 590                         return;
 591
 592                 if ( is_object( $nt ) )
 593                         $ns = $nt->getNamespace();
 594
 595                 $variants = $this->autoConvertToAllVariants($link);
 596                 if($variants == false) //give up
 597                         return;
 598
 599                 $titles = array();
 600
 601                 foreach( $variants as $v ) {
 602                         if($v != $link){
 603                                 $varnt = Title::newFromText( $v, $ns );
 604                                 if(!is_null($varnt)){
 605                                         $linkBatch->addObj($varnt);
 606                                         $titles[]=$varnt;
 607                                 }
 608                         }
 609                 }
 610
 611                 // fetch all variants in single query
 612                 $linkBatch->execute();
 613
 614                 foreach( $titles as $varnt ) {
 615                         if( $varnt->getArticleID() > 0 ) {
 616                                 $nt = $varnt;
 617                                 $link = $varnt->getText();
 618                                 break;
 619                         }
 620                 }
 621         }
 622
 623     /**
 624          * returns language specific hash options
 625          *
 626          * @public
 627          */
 628         function getExtraHashOptions() {
 629                 $variant = $this->getPreferredVariant();
 630                 return '!' . $variant ;
 631         }
 632
 633     /**
 634          * get title text as defined in the body of the article text
 635          *
 636          * @public
 637          */
 638         function getParsedTitle() {
 639                 return $this->mTitleDisplay;
 640         }
 641
 642         /**
 643          * Load default conversion tables
 644          * This method must be implemented in derived class
 645          *
 646          * @private
 647          */
 648         function loadDefaultTables() {
 649                 $name = get_class($this);
 650                 wfDie("Must implement loadDefaultTables() method in class $name");
 651         }
 652
 653         /**
 654          * load conversion tables either from the cache or the disk
 655          * @private
 656          */
 657         function loadTables($fromcache=true) {
 658                 global $wgMemc;
 659                 if( $this->mTablesLoaded )
 660                         return;
 661                 wfProfileIn( __METHOD__ );
 662                 $this->mTablesLoaded = true;
 663                 $this->mTables = false;
 664                 if($fromcache) {
 665                         wfProfileIn( __METHOD__.'-cache' );
 666                         $this->mTables = $wgMemc->get( $this->mCacheKey );
 667                         wfProfileOut( __METHOD__.'-cache' );
 668                 }
 669                 if ( !$this->mTables || !isset( $this->mTables[self::CACHE_VERSION_KEY] ) ) {
 670                         wfProfileIn( __METHOD__.'-recache' );
 671                         // not in cache, or we need a fresh reload.
 672                         // we will first load the default tables
 673                         // then update them using things in MediaWiki:Zhconversiontable/*
 674                         $this->loadDefaultTables();
 675                         foreach($this->mVariants as $var) {
 676                                 $cached = $this->parseCachedTable($var);
 677                                 $this->mTables[$var]->mergeArray($cached);
 678                         }
 679
 680                         $this->postLoadTables();
 681                         $this->mTables[self::CACHE_VERSION_KEY] = true;
 682
 683                         $wgMemc->set($this->mCacheKey, $this->mTables, 43200);
 684                         wfProfileOut( __METHOD__.'-recache' );
 685                 }
 686                 wfProfileOut( __METHOD__ );
 687         }
 688
 689     /**
 690          * Hook for post processig after conversion tables are loaded
 691          *
 692          */
 693         function postLoadTables() {}
 694
 695     /**
 696          * Reload the conversion tables
 697          *
 698          * @private
 699          */
 700         function reloadTables() {
 701                 if($this->mTables)
 702                         unset($this->mTables);
 703                 $this->mTablesLoaded = false;
 704                 $this->loadTables(false);
 705         }
 706
 707
 708         /**
 709          * parse the conversion table stored in the cache
 710          *
 711          * the tables should be in blocks of the following form:
 712          *              -{
 713          *                      word => word ;
 714          *                      word => word ;
 715          *                      ...
 716          *              }-
 717          *
 718          *      to make the tables more manageable, subpages are allowed
 719          *      and will be parsed recursively if $recursive=true
 720          *
 721          */
 722         function parseCachedTable($code, $subpage='', $recursive=true) {
 723                 global $wgMessageCache;
 724                 static $parsed = array();
 725
 726                 if(!is_object($wgMessageCache))
 727                         return array();
 728
 729                 $key = 'Conversiontable/'.$code;
 730                 if($subpage)
 731                         $key .= '/' . $subpage;
 732
 733                 if(array_key_exists($key, $parsed))
 734                         return array();
 735
 736                 if ( strpos( $code, '/' ) === false ) {
 737                         $txt = $wgMessageCache->get( 'Conversiontable', true, $code );
 738                 } else {
 739                         $title = Title::makeTitleSafe( NS_MEDIAWIKI, "Conversiontable/$code" );
 740                         if ( $title && $title->exists() ) {
 741                                 $article = new Article( $title );
 742                                 $txt = $article->getContents();
 743                         } else {
 744                                 $txt = '';
 745                         }
 746                 }
 747
 748                 // get all subpage links of the form
 749                 // [[MediaWiki:conversiontable/zh-xx/...|...]]
 750                 $linkhead = $this->mLangObj->getNsText(NS_MEDIAWIKI) . ':Conversiontable';
 751                 $subs = explode('[[', $txt);
 752                 $sublinks = array();
 753                 foreach( $subs as $sub ) {
 754                         $link = explode(']]', $sub, 2);
 755                         if(count($link) != 2)
 756                                 continue;
 757                         $b = explode('|', $link[0]);
 758                         $b = explode('/', trim($b[0]), 3);
 759                         if(count($b)==3)
 760                                 $sublink = $b[2];
 761                         else
 762                                 $sublink = '';
 763
 764                         if($b[0] == $linkhead && $b[1] == $code) {
 765                                 $sublinks[] = $sublink;
 766                         }
 767                 }
 768
 769
 770                 // parse the mappings in this page
 771                 $blocks = explode($this->mMarkup['begin'], $txt);
 772                 array_shift($blocks);
 773                 $ret = array();
 774                 foreach($blocks as $block) {
 775                         $mappings = explode($this->mMarkup['end'], $block, 2);
 776                         $stripped = str_replace(array("'", '"', '*','#'), '', $mappings[0]);
 777                         $table = explode( ';', $stripped );
 778                         foreach( $table as $t ) {
 779                                 $m = explode( '=>', $t );
 780                                 if( count( $m ) != 2)
 781                                         continue;
 782                                 // trim any trailling comments starting with '//'
 783                                 $tt = explode('//', $m[1], 2);
 784                                 $ret[trim($m[0])] = trim($tt[0]);
 785                         }
 786                 }
 787                 $parsed[$key] = true;
 788
 789
 790                 // recursively parse the subpages
 791                 if($recursive) {
 792                         foreach($sublinks as $link) {
 793                                 $s = $this->parseCachedTable($code, $link, $recursive);
 794                                 $ret = array_merge($ret, $s);
 795                         }
 796                 }
 797
 798                 if ($this->mUcfirst) {
 799                         foreach ($ret as $k => $v) {
 800                                 $ret[Language::ucfirst($k)] = Language::ucfirst($v);
 801                         }
 802                 }
 803                 return $ret;
 804         }
 805
 806         /**
 807          * Enclose a string with the "no conversion" tag. This is used by
 808          * various functions in the Parser
 809          *
 810          * @param string $text text to be tagged for no conversion
 811          * @return string the tagged text
 812          * @public
 813          */
 814         function markNoConversion($text, $noParse=false) {
 815                 # don't mark if already marked
 816                 if(strpos($text, $this->mMarkup['begin']) ||
 817                    strpos($text, $this->mMarkup['end']))
 818                         return $text;
 819
 820                 $ret = $this->mMarkup['begin'] .'R|'. $text . $this->mMarkup['end'];
 821                 return $ret;
 822         }
 823
 824         /**
 825          * convert the sorting key for category links. this should make different
 826          * keys that are variants of each other map to the same key
 827          */
 828         function convertCategoryKey( $key ) {
 829                 return $key;
 830         }
 831         /**
 832          * hook to refresh the cache of conversion tables when
 833          * MediaWiki:conversiontable* is updated
 834          * @private
 835          */
 836         function OnArticleSaveComplete($article, $user, $text, $summary, $isminor, $iswatch, $section, $flags, $revision) {
 837                 $titleobj = $article->getTitle();
 838                 if($titleobj->getNamespace() == NS_MEDIAWIKI) {
 839                         $title = $titleobj->getDBkey();
 840                         $t = explode('/', $title, 3);
 841                         $c = count($t);
 842                         if( $c > 1 && $t[0] == 'Conversiontable' ) {
 843                                 if(in_array($t[1], $this->mVariants)) {
 844                                         $this->reloadTables();
 845                                 }
 846                         }
 847                 }
 848                 return true;
 849         }
 850
 851         /**
 852          * Armour rendered math against conversion
 853          * Wrap math into rawoutput -{R| math }- syntax
 854          * @public
 855          */
 856         function armourMath($text){
 857                 // we need to convert '-{' and '}-' to '-&#123;' and '&#125;-'
 858                 // to avoid a unwanted '}-' appeared after the math-image.
 859                 $text = strtr( $text, array('-{' => '-&#123;', '}-' => '&#125;-') );
 860                 $ret = $this->mMarkup['begin'] . 'R|' . $text . $this->mMarkup['end'];
 861                 return $ret;
 862         }
 863 }
 864
 865 /**
 866  * parser for rules of language conversion , parse rules in -{ }- tag
 867  * @ingroup Language
 868  * @author fdcn <fdcn64@gmail.com>, PhiLiP <philip.npc@gmail.com>
 869  */
 870 class ConverterRule {
 871         var $mText; // original text in -{text}-
 872         var $mConverter; // LanguageConverter object
 873         var $mManualCodeError='<strong class="error">code error!</strong>';
 874         var $mRuleDisplay = '',$mRuleTitle=false;
 875         var $mRules = '';// string : the text of the rules
 876         var $mRulesAction = 'none';
 877         var $mFlags = array();
 878         var $mConvTable = array();
 879         var $mBidtable = array();// array of the translation in each variant
 880         var $mUnidtable = array();// array of the translation in each variant
 881
 882         /**
 883          * Constructor
 884          *
 885          * @param string $text the text between -{ and }-
 886          * @param object $converter a  LanguageConverter object
 887          * @access public
 888          */
 889         function __construct($text,$converter){
 890                 $this->mText = $text;
 891                 $this->mConverter=$converter;
 892                 foreach($converter->mVariants as $v){
 893                         $this->mConvTable[$v]=array();
 894                 }
 895         }
 896
 897         /**
 898          * check if variants array in convert array
 899          *
 900          * @param string $variant Variant language code
 901          * @return string Translated text
 902          * @public
 903          */
 904         function getTextInBidtable($variants){
 905                 if(is_string($variants)){ $variants=array($variants); }
 906                 if(!is_array($variants)) return false;
 907                 foreach ($variants as $variant){
 908                         if(array_key_exists($variant, $this->mBidtable)){
 909                                 return $this->mBidtable[$variant];
 910                         }
 911                 }
 912                 return false;
 913         }
 914
 915         /**
 916          * Parse flags with syntax -{FLAG| ... }-
 917          * @private
 918          */
 919         function parseFlags(){
 920                 $text = $this->mText;
 921                 if(strlen($text) < 2 ) {
 922                         $this->mFlags = array( 'R' );
 923                         $this->mRules = $text;
 924                         return;
 925                 }
 926
 927                 $flags = array();
 928                 $markup = $this->mConverter->mMarkup;
 929                 $validFlags = $this->mConverter->mFlags;
 930                 $variants = $this->mConverter->mVariants;
 931
 932                 $tt = explode($markup['flagsep'], $text, 2);
 933                 if(count($tt) == 2) {
 934                         $f = explode($markup['varsep'], $tt[0]);
 935                         foreach($f as $ff) {
 936                                 $ff = trim($ff);
 937                                 if(array_key_exists($ff, $validFlags) &&
 938                                                         !in_array($validFlags[$ff], $flags))
 939                                         $flags[] = $validFlags[$ff];
 940                         }
 941                         $rules = $tt[1];
 942                 } else {
 943                         $rules = $text;
 944                 }
 945
 946                 //check flags
 947                 if( in_array('R',$flags) ){
 948                         $flags = array('R');// remove other flags
 949                 } elseif ( in_array('N',$flags) ){
 950                         $flags = array('N');// remove other flags
 951                 } elseif ( in_array('-',$flags) ){
 952                         $flags = array('-');// remove other flags
 953                 } elseif (count($flags)==1 && $flags[0]=='T'){
 954                         $flags[]='H';
 955                 } elseif ( in_array('H',$flags) ){
 956                         // replace A flag, and remove other flags except T
 957                         $temp=array('+','H');
 958                         if(in_array('T',$flags)) $temp[] = 'T';
 959                         if(in_array('D',$flags)) $temp[] = 'D';
 960                         $flags = $temp;
 961                 } else {
 962                         if ( in_array('A',$flags) ) {
 963                                 $flags[]='+';
 964                                 $flags[]='S';
 965                         }
 966                         if ( in_array('D',$flags) )
 967                                 $flags=array_diff($flags,array('S'));
 968                         $flags_temp = array();
 969                         foreach ($variants as $variant) {
 970                                 // try to find flags like "zh-hans", "zh-hant"
 971                                 // allow syntaxes like "-{zh-hans;zh-hant|XXXX}-"
 972                                 if ( in_array($variant, $flags) )
 973                                         $flags_temp[] = $variant;
 974                         }
 975                         if ( count($flags_temp) !== 0 )
 976                                 $flags = $flags_temp;
 977                 }
 978                 if ( count($flags) == 0 )
 979                         $flags = array('S');
 980                 $this->mRules=$rules;
 981                 $this->mFlags=$flags;
 982         }
 983
 984         /**
 985          * generate conversion table
 986          * @private
 987          */
 988         function parseRules() {
 989                 $rules = $this->mRules;
 990                 $flags = $this->mFlags;
 991                 $bidtable = array();
 992                 $unidtable = array();
 993                 $markup = $this->mConverter->mMarkup;
 994                 $variants = $this->mConverter->mVariants;
 995
 996                 // varsep_pattern for preg_split:
 997                 // text should be splited by ";" only if a valid variant
 998                 // name exist after the markup, for example:
 999                 //  -{zh-hans:<span style="font-size:120%;">xxx</span>;zh-hant:<span style="font-size:120%;">yyy</span>;}-
1000                 // we should split it as:
1001                 //  array(
1002                 //        [0] => 'zh-hans:<span style="font-size:120%;">xxx</span>'
1003                 //        [1] => 'zh-hant:<span style="font-size:120%;">yyy</span>'
1004                 //        [2] => ''
1005                 //       )
1006                 $varsep_pattern = '/' . $markup['varsep'] . '\s*' . '(?=';
1007                 foreach( $variants as $variant ) {
1008                         $varsep_pattern .= $variant . '\s*' . $markup['codesep'] . '|'; // zh-hans:xxx;zh-hant:yyy
1009                         $varsep_pattern .= '[^;]*?' . $markup['unidsep'] . '\s*' . $variant
1010                                                         . '\s*' . $markup['codesep'] . '|'; // xxx=>zh-hans:yyy; xxx=>zh-hant:zzz
1011                 }
1012                 $varsep_pattern .= '\s*$)/';
1013
1014                 $choice = preg_split($varsep_pattern, $rules);
1015
1016                 foreach( $choice as $c ) {
1017                         $v  = explode($markup['codesep'], $c, 2);
1018                         if( count($v) != 2 )
1019                                 continue;// syntax error, skip
1020                         $to = trim($v[1]);
1021                         $v  = trim($v[0]);
1022                         $u  = explode($markup['unidsep'], $v);
1023                         // if $to is empty, strtr() could return a wrong result
1024                         if( count($u) == 1 && $to && in_array( $v, $variants ) ) {
1025                                 $bidtable[$v] = $to;
1026                         } else if(count($u) == 2){
1027                                 $from = trim($u[0]);
1028                                 $v    = trim($u[1]);
1029                                 if( array_key_exists( $v, $unidtable ) && !is_array( $unidtable[$v] )
1030                                         && $to && in_array( $v, $variants ) )
1031                                         $unidtable[$v] = array( $from=>$to );
1032                                 elseif ( $to && in_array( $v, $variants ) )
1033                                         $unidtable[$v][$from] = $to;
1034                         }
1035                         // syntax error, pass
1036                         if ( !array_key_exists( $v, $this->mConverter->mVariantNames ) ){
1037                                 $bidtable = array();
1038                                 $unidtable = array();
1039                                 break;
1040                         }
1041                 }
1042                 $this->mBidtable = $bidtable;
1043                 $this->mUnidtable = $unidtable;
1044         }
1045
1046         /**
1047          * @private
1048          */
1049         function getRulesDesc(){
1050                 $codesep = $this->mConverter->mDescCodeSep;
1051                 $varsep = $this->mConverter->mDescVarSep;
1052                 $text='';
1053                 foreach($this->mBidtable as $k => $v)
1054                         $text .= $this->mConverter->mVariantNames[$k]."$codesep$v$varsep";
1055                 foreach($this->mUnidtable as $k => $a)
1056                         foreach($a as $from=>$to)
1057                                 $text.=$from.'⇒'.$this->mConverter->mVariantNames[$k]."$codesep$to$varsep";
1058                 return $text;
1059         }
1060
1061         /**
1062          *  Parse rules conversion
1063          * @private
1064          */
1065         function getRuleConvertedStr($variant,$doConvert){
1066                 $bidtable = $this->mBidtable;
1067                 $unidtable = $this->mUnidtable;
1068
1069                 if( count($bidtable) + count($unidtable) == 0 ){
1070                         return $this->mRules;
1071                 } elseif ($doConvert){// the text converted
1072                         // display current variant in bidirectional array
1073                         $disp = $this->getTextInBidtable($variant);
1074                         // or display current variant in fallbacks
1075                         if(!$disp)
1076                                 $disp = $this->getTextInBidtable(
1077                                                 $this->mConverter->getVariantFallbacks($variant));
1078                         // or display current variant in unidirectional array
1079                         if(!$disp && array_key_exists($variant,$unidtable)){
1080                                 $disp = array_values($unidtable[$variant]);
1081                                 $disp = $disp[0];
1082                         }
1083                         // or display frist text under disable manual convert
1084                         if(!$disp && $this->mConverter->mManualLevel[$variant]=='disable') {
1085                                 if(count($bidtable)>0){
1086                                         $disp = array_values($bidtable);
1087                                         $disp = $disp[0];
1088                                 } else {
1089                                         $disp = array_values($unidtable);
1090                                         $disp = array_values($disp[0]);
1091                                         $disp = $disp[0];
1092                                 }
1093                         }
1094                         return $disp;
1095                 } else {// no convert
1096                         return $this->mRules;
1097                 }
1098         }
1099
1100         /**
1101          * generate conversion table for all text
1102          * @private
1103          */
1104         function generateConvTable(){
1105                 $flags = $this->mFlags;
1106                 $bidtable = $this->mBidtable;
1107                 $unidtable = $this->mUnidtable;
1108                 $manLevel = $this->mConverter->mManualLevel;
1109
1110                 $vmarked=array();
1111                 foreach($this->mConverter->mVariants as $v) {
1112                         /* for bidirectional array
1113                                 fill in the missing variants, if any,
1114                                 with fallbacks */
1115                         if(!array_key_exists($v, $bidtable)) {
1116                                 $variantFallbacks = $this->mConverter->getVariantFallbacks($v);
1117                                 $vf = $this->getTextInBidtable($variantFallbacks);
1118                                 if($vf) $bidtable[$v] = $vf;
1119                         }
1120
1121                         if(array_key_exists($v,$bidtable)){
1122                                 foreach($vmarked as $vo){
1123                                         // use syntax: -{A|zh:WordZh;zh-tw:WordTw}-
1124                                         // or -{H|zh:WordZh;zh-tw:WordTw}- or -{-|zh:WordZh;zh-tw:WordTw}-
1125                                         // to introduce a custom mapping between
1126                                         // words WordZh and WordTw in the whole text
1127                                         if($manLevel[$v]=='bidirectional'){
1128                                                 $this->mConvTable[$v][$bidtable[$vo]]=$bidtable[$v];
1129                                         }
1130                                         if($manLevel[$vo]=='bidirectional'){
1131                                                 $this->mConvTable[$vo][$bidtable[$v]]=$bidtable[$vo];
1132                                         }
1133                                 }
1134                                 $vmarked[]=$v;
1135                         }
1136                         /*for unidirectional array
1137                                 fill to convert tables */
1138                         $allow_unid = $manLevel[$v]=='bidirectional'
1139                                         || $manLevel[$v]=='unidirectional';
1140                         if($allow_unid && array_key_exists($v,$unidtable)){
1141                                 $ct=$this->mConvTable[$v];
1142                                 $this->mConvTable[$v] = array_merge($ct,$unidtable[$v]);
1143                         }
1144                 }
1145         }
1146
1147         /**
1148          * Parse rules and flags
1149          * @public
1150          */
1151         function parse($variant){
1152                 if(!$variant)
1153                         $variant = $this->mConverter->getPreferredVariant();
1154
1155                 $variants = $this->mConverter->mVariants;
1156                 $this->parseFlags();
1157                 $flags = $this->mFlags;
1158
1159                 // convert to specified variant
1160                 // syntax: -{zh-hans;zh-hant[;...]|<text to convert>}-
1161                 if( count( array_diff( $flags, $variants ) ) == 0 and count( $flags ) != 0 ) {
1162                         if ( in_array( $variant, $flags ) ) // check if current variant in flags
1163                                 // then convert <text to convert> to current language
1164                                 $this->mRules = $this->mConverter->autoConvert( $this->mRules, $variant );
1165                         else { // if current variant no in flags,
1166                                    // then we check its fallback variants.
1167                                 $variantFallbacks = $this->mConverter->getVariantFallbacks($variant);
1168                                 foreach ( $variantFallbacks as $variantFallback ) {
1169                                         // if current variant's fallback exist in flags
1170                                         if ( in_array( $variantFallback, $flags ) ) {
1171                                                 // then convert <text to convert> to fallback language
1172                                                 $this->mRules = $this->mConverter->autoConvert( $this->mRules, $variantFallback );
1173                                                 break;
1174                                         }
1175                                 }
1176                         }
1177                         $this->mFlags = $flags = array('R');
1178                 }
1179
1180                 if( !in_array( 'R', $flags ) || !in_array( 'N', $flags ) ) {
1181                         // decode => HTML entities modified by Sanitizer::removeHTMLtags
1182                         $this->mRules = str_replace('=&gt;','=>',$this->mRules);
1183
1184                         $this->parseRules();
1185                 }
1186                 $rules = $this->mRules;
1187
1188                 if( count( $this->mBidtable ) == 0 && count( $this->mUnidtable ) == 0 ){
1189                         if(in_array('+',$flags) || in_array('-',$flags))
1190                                 // fill all variants if text in -{A/H/-|text} without rules
1191                                 foreach($this->mConverter->mVariants as $v)
1192                                         $this->mBidtable[$v] = $rules;
1193                         elseif (!in_array('N',$flags) && !in_array('T',$flags) )
1194                                 $this->mFlags = $flags = array('R');
1195                 }
1196
1197                 if( in_array('R',$flags) ) {
1198                         // if we don't do content convert, still strip the -{}- tags
1199                         $this->mRuleDisplay = $rules;
1200                 } elseif ( in_array('N',$flags) ){
1201                         // proces N flag: output current variant name
1202                         $this->mRuleDisplay = $this->mConverter->mVariantNames[trim($rules)];
1203                 } elseif ( in_array('D',$flags) ){
1204                         // proces D flag: output rules description
1205                         $this->mRuleDisplay = $this->getRulesDesc();
1206                 } elseif ( in_array('H',$flags) || in_array('-',$flags) ) {
1207                         // proces H,- flag or T only: output nothing
1208                         $this->mRuleDisplay = '';
1209                 } elseif ( in_array('S',$flags) ){
1210                         $this->mRuleDisplay = $this->getRuleConvertedStr($variant,
1211                                                         $this->mConverter->mDoContentConvert);
1212                 } else {
1213                         $this->mRuleDisplay= $this->mManualCodeError;
1214                 }
1215                 // proces T flag
1216                 if ( in_array('T',$flags) ) {
1217                         $this->mRuleTitle = $this->getRuleConvertedStr($variant,
1218                                                         $this->mConverter->mDoTitleConvert);
1219                 }
1220
1221                 if (in_array('-', $flags))
1222                         $this->mRulesAction='remove';
1223                 if (in_array('+', $flags))
1224                         $this->mRulesAction='add';
1225
1226                 $this->generateConvTable();
1227         }
1228
1229         /**
1230          * @public
1231          */
1232         function hasRules(){
1233                 // TODO:
1234         }
1235
1236         /**
1237          * get display text on markup -{...}-
1238          * @public
1239          */
1240         function getDisplay(){
1241                 return $this->mRuleDisplay;
1242         }
1243         /**
1244          * get converted title
1245          * @public
1246          */
1247         function getTitle(){
1248                 return $this->mRuleTitle;
1249         }
1250
1251         /**
1252          * return how deal with conversion rules
1253          * @public
1254          */
1255         function getRulesAction(){
1256                 return $this->mRulesAction;
1257         }
1258
1259         /**
1260          * get conversion table ( bidirectional and unidirectional conversion table )
1261          * @public
1262          */
1263         function getConvTable(){
1264                 return $this->mConvTable;
1265         }
1266
1267         /**
1268          * get conversion rules string
1269          * @public
1270          */
1271         function getRules(){
1272                 return $this->mRules;
1273         }
1274
1275         /**
1276          * get conversion flags
1277          * @public
1278          */
1279         function getFlags(){
1280                 return $this->mFlags;
1281         }
1282 }