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                 $this->mVariants = $variants;
  58                 $this->mVariantFallbacks = $variantfallbacks;
  59                 global $wgLanguageNames;
  60                 $this->mVariantNames = $wgLanguageNames;
  61                 $this->mCacheKey = wfMemcKey( 'conversiontables', $maincode );
  62                 $m = array(
  63                         'begin'=>'-{',
  64                         'flagsep'=>'|',
  65                         'unidsep'=>'=>', //for unidirectional conversion
  66                         'codesep'=>':',
  67                         'varsep'=>';',
  68                         'end'=>'}-'
  69                 );
  70                 $this->mMarkup = array_merge($m, $markup);
  71                 $f = array(
  72                         // 'S' show converted text
  73                         // '+' add rules for alltext
  74                         // 'E' the gave flags is error
  75                         // these flags above are reserved for program
  76                         'A'=>'A',       // add rule for convert code (all text convert)
  77                         'T'=>'T',       // title convert
  78                         'R'=>'R',       // raw content
  79                         'D'=>'D',       // convert description (subclass implement)
  80                         '-'=>'-',       // remove convert (not implement)
  81                         'H'=>'H',       // add rule for convert code (but no display in placed code )
  82                         'N'=>'N'        // current variant name
  83                 );
  84                 $this->mFlags = array_merge($f, $flags);
  85                 foreach( $this->mVariants as $v) {
  86                         $this->mManualLevel[$v]=array_key_exists($v,$manualLevel)
  87                                                                 ?$manualLevel[$v]
  88                                                                 :'bidirectional';
  89                         $this->mNamespaceTables[$v] = array();
  90                         $this->mFlags[$v] = $v;
  91                 }
  92         }
  93
  94         /**
  95          * @public
  96          */
  97         function getVariants() {
  98                 return $this->mVariants;
  99         }
 100
 101         /**
 102          * in case some variant is not defined in the markup, we need
 103          * to have some fallback. for example, in zh, normally people
 104          * will define zh-hans and zh-hant, but less so for zh-sg or zh-hk.
 105          * when zh-sg is preferred but not defined, we will pick zh-hans
 106          * in this case. right now this is only used by zh.
 107          *
 108          * @param string $v the language code of the variant
 109          * @return string array the code of the fallback language or false if there is no fallback
 110          * @public
 111          */
 112         function getVariantFallbacks($v) {
 113                 if( isset( $this->mVariantFallbacks[$v] ) ) {
 114                         return $this->mVariantFallbacks[$v];
 115                 }
 116                 return $this->mMainLanguageCode;
 117         }
 118
 119         /**
 120          * get preferred language variants.
 121          * @param boolean $fromUser Get it from $wgUser's preferences
 122          * @return string the preferred language code
 123          * @public
 124          */
 125         function getPreferredVariant( $fromUser = true ) {
 126                 global $wgUser, $wgRequest, $wgVariantArticlePath, $wgDefaultLanguageVariant;
 127
 128                 if($this->mPreferredVariant)
 129                         return $this->mPreferredVariant;
 130
 131                 // figure out user lang without constructing wgLang to avoid infinite recursion
 132                 if( $fromUser )
 133                         $defaultUserLang = $wgUser->getOption( 'language' );
 134                 else
 135                         $defaultUserLang = $this->mMainLanguageCode;
 136                 $userLang = $wgRequest->getVal( 'uselang', $defaultUserLang );
 137                 // see if interface language is same as content, if not, prevent conversion
 138                 if( ! in_array( $userLang, $this->mVariants ) ){
 139                         $this->mPreferredVariant = $this->mMainLanguageCode; // no conversion
 140                         return $this->mPreferredVariant;
 141                 }
 142
 143                 // see if the preference is set in the request
 144                 $req = $wgRequest->getText( 'variant' );
 145                 if( in_array( $req, $this->mVariants ) ) {
 146                         $this->mPreferredVariant = $req;
 147                         return $req;
 148                 }
 149
 150                 // check the syntax /code/ArticleTitle
 151                 if($wgVariantArticlePath!=false && isset($_SERVER['SCRIPT_NAME'])){
 152                         // Note: SCRIPT_NAME probably won't hold the correct value if PHP is run as CGI
 153                         // (it will hold path to php.cgi binary), and might not exist on some very old PHP installations
 154                         $scriptBase = basename( $_SERVER['SCRIPT_NAME'] );
 155                         if(in_array($scriptBase,$this->mVariants)){
 156                                 $this->mPreferredVariant = $scriptBase;
 157                                 return $this->mPreferredVariant;
 158                         }
 159                 }
 160
 161                 // get language variant preference from logged in users
 162                 // Don't call this on stub objects because that causes infinite
 163                 // recursion during initialisation
 164                 if( $fromUser && $wgUser->isLoggedIn() )  {
 165                         $this->mPreferredVariant = $wgUser->getOption('variant');
 166                         return $this->mPreferredVariant;
 167                 }
 168
 169                 // see if default variant is globaly set
 170                 if($wgDefaultLanguageVariant != false  &&  in_array( $wgDefaultLanguageVariant, $this->mVariants )){
 171                         $this->mPreferredVariant = $wgDefaultLanguageVariant;
 172                         return $this->mPreferredVariant;
 173                 }
 174
 175                 if( !$this->mPreferredVariant ) {
 176                         // see if some supported language variant is set in the
 177                         // http header, but we don't set the mPreferredVariant
 178                         // variable in case this is called before the user's
 179                         // preference is loaded
 180                         if( array_key_exists( 'HTTP_ACCEPT_LANGUAGE', $_SERVER ) ) {
 181                                 $acceptLanguage = strtolower( $_SERVER['HTTP_ACCEPT_LANGUAGE'] );
 182
 183                                 // explode by comma
 184                                 $result = explode(',', $acceptLanguage);
 185
 186                                 $languages = array();
 187
 188                                 foreach( $result as $elem ) {
 189                                         // if $elem likes 'zh-cn;q=0.9'
 190                                         if(($posi = strpos( $elem, ';' )) !== false ) {
 191                                                 // get the real language code likes 'zh-cn'
 192                                                 $languages[] = substr( $elem, 0, $posi );
 193                                         }
 194                                         else {
 195                                                 $languages[] = $elem;
 196                                         }
 197                                 }
 198
 199                                 $fallback_languages = array();
 200                                 foreach( $languages as $language ) {
 201                                         // strip whitespace
 202                                         $language = trim( $language );
 203                                         if( in_array( $language, $this->mVariants ) ) {
 204                                                 return $language;
 205                                         }
 206                                         else {
 207                                                 // To see if there are fallbacks of current language.
 208                                                 // We record these fallback variants, and process
 209                                                 // them later.
 210                                                 $fallbacks = $this->getVariantFallbacks( $language );
 211                                                 if( is_string( $fallbacks ) )
 212                                                         $fallback_languages[] = $fallbacks;
 213                                                 elseif( is_array( $fallbacks ) )
 214                                                         $fallback_languages = array_merge( $fallback_languages, $fallbacks );
 215                                         }
 216                                 }
 217
 218                                 // process fallback languages now
 219                                 $fallback_languages = array_unique( $fallback_languages );
 220                                 foreach( $fallback_languages as $language ) {
 221                                         if( in_array( $language, $this->mVariants ) ) {
 222                                                 return $language;
 223                                         }
 224                                 }
 225                         }
 226                 }
 227
 228                 return $this->mMainLanguageCode;
 229         }
 230
 231         /**
 232          * caption convert, base on preg_replace_callback
 233          *
 234          * to convert text in "title" or "alt", like '<img alt="text" ... '
 235          * or '<span title="text" ... '
 236          *
 237          * @return string like ' alt="yyyy"' or ' title="yyyy"'
 238          * @private
 239          */
 240         function captionConvert( $matches ) {
 241                 $toVariant = $this->getPreferredVariant();
 242                 $title = $matches[1];
 243                 $text  = $matches[2];
 244                 // we convert captions except URL
 245                 if( !strpos( $text, '://' ) )
 246                         $text = $this->translate($text, $toVariant);
 247                 return " $title=\"$text\"";
 248         }
 249
 250         /**
 251          * dictionary-based conversion
 252          *
 253          * @param string $text the text to be converted
 254          * @param string $toVariant the target language code
 255          * @return string the converted text
 256          * @private
 257          */
 258         function autoConvert($text, $toVariant=false) {
 259                 $fname="LanguageConverter::autoConvert";
 260
 261                 wfProfileIn( $fname );
 262
 263                 if(!$this->mTablesLoaded)
 264                         $this->loadTables();
 265
 266                 if(!$toVariant)
 267                         $toVariant = $this->getPreferredVariant();
 268                 if(!in_array($toVariant, $this->mVariants))
 269                         return $text;
 270
 271                 /* we convert everything except:
 272                    1. html markups (anything between < and >)
 273                    2. html entities
 274                    3. place holders created by the parser
 275                 */
 276                 global $wgParser;
 277                 if (isset($wgParser) && $wgParser->UniqPrefix()!=''){
 278                         $marker = '|' . $wgParser->UniqPrefix() . '[\-a-zA-Z0-9]+';
 279                 } else
 280                         $marker = "";
 281
 282                 // this one is needed when the text is inside an html markup
 283                 $htmlfix = '|<[^>]+$|^[^<>]*>';
 284
 285                 // disable convert to variants between <code></code> tags
 286                 $codefix = '<code>.+?<\/code>|';
 287                 // disable convertsion of <script type="text/javascript"> ... </script>
 288                 $scriptfix = '<script.*?>.*?<\/script>|';
 289                 // disable conversion of <pre xxxx> ... </pre>
 290                 $prefix = '<pre.*?>.*?<\/pre>|';
 291
 292                 $reg = '/'.$codefix . $scriptfix . $prefix . '<[^>]+>|&[a-zA-Z#][a-z0-9]+;' . $marker . $htmlfix . '/s';
 293
 294                 $matches = preg_split($reg, $text, -1, PREG_SPLIT_OFFSET_CAPTURE);
 295
 296                 $m = array_shift($matches);
 297
 298                 $ret = $this->translate($m[0], $toVariant);
 299                 $mstart = $m[1]+strlen($m[0]);
 300
 301                 // enable convertsion of '<img alt="xxxx" ... ' or '<span title="xxxx" ... '
 302                 $captionpattern  = '/\s(title|alt)\s*=\s*"([\s\S]*?)"/';
 303                 foreach($matches as $m) {
 304                         $mark = substr($text, $mstart, $m[1]-$mstart);
 305                         $mark = preg_replace_callback($captionpattern, array(&$this, 'captionConvert'), $mark);
 306                         $ret .= $mark;
 307                         $ret .= $this->translate($m[0], $toVariant);
 308                         $mstart = $m[1] + strlen($m[0]);
 309                 }
 310                 wfProfileOut( $fname );
 311                 return $ret;
 312         }
 313
 314         /**
 315          * Translate a string to a variant
 316          * Doesn't process markup or do any of that other stuff, for that use convert()
 317          *
 318          * @param string $text Text to convert
 319          * @param string $variant Variant language code
 320          * @return string Translated text
 321          * @private
 322          */
 323         function translate( $text, $variant ) {
 324                 wfProfileIn( __METHOD__ );
 325                 if( !$this->mTablesLoaded )
 326                         $this->loadTables();
 327                 $text = $this->mTables[$variant]->replace( $text );
 328                 wfProfileOut( __METHOD__ );
 329                 return $text;
 330         }
 331
 332         /**
 333          * convert text to all supported variants
 334          *
 335          * @param string $text the text to be converted
 336          * @return array of string
 337          * @public
 338          */
 339         function autoConvertToAllVariants($text) {
 340                 $fname="LanguageConverter::autoConvertToAllVariants";
 341                 wfProfileIn( $fname );
 342                 if( !$this->mTablesLoaded )
 343                         $this->loadTables();
 344
 345                 $ret = array();
 346                 foreach($this->mVariants as $variant) {
 347                         $ret[$variant] = $this->translate($text, $variant);
 348                 }
 349
 350                 wfProfileOut( $fname );
 351                 return $ret;
 352         }
 353
 354         /**
 355          * convert link text to all supported variants
 356          *
 357          * @param string $text the text to be converted
 358          * @return array of string
 359          * @public
 360          */
 361         function convertLinkToAllVariants($text) {
 362                 if( !$this->mTablesLoaded )
 363                         $this->loadTables();
 364
 365                 $ret = array();
 366                 $tarray = explode($this->mMarkup['begin'], $text);
 367                 $tfirst = array_shift($tarray);
 368
 369                 foreach($this->mVariants as $variant)
 370                         $ret[$variant] = $this->translate($tfirst,$variant);
 371
 372                 foreach($tarray as $txt) {
 373                         $marked = explode($this->mMarkup['end'], $txt, 2);
 374
 375                         foreach($this->mVariants as $variant){
 376                                 $ret[$variant] .= $this->mMarkup['begin'].$marked[0].$this->mMarkup['end'];
 377                                 if(array_key_exists(1, $marked))
 378                                         $ret[$variant] .= $this->translate($marked[1],$variant);
 379                         }
 380
 381                 }
 382
 383                 return $ret;
 384         }
 385
 386         /**
 387          * prepare manual conversion table
 388          * @private
 389          */
 390         function applyManualConv( $convRule ){
 391                 // use syntax -{T|zh:TitleZh;zh-tw:TitleTw}- for custom conversion in title
 392                 $title = $convRule->getTitle();
 393                 if( $title ){
 394                         $this->mTitleFromFlag = true;
 395                         $this->mTitleDisplay =  $title;
 396                 }
 397
 398                 //apply manual conversion table to global table
 399                 $convTable = $convRule->getConvTable();
 400                 $action = $convRule->getRulesAction();
 401                 foreach( $convTable as $variant => $pair ) {
 402                         if( !in_array( $variant, $this->mVariants ) )continue;
 403                         if( $action=="add" ) {
 404                                 foreach( $pair as $from => $to ) {
 405                                         // to ensure that $from and $to not be left blank
 406                                         // so $this->translate() could always return a string
 407                                         if ( $from || $to )
 408                                                 // more efficient than array_merge(), about 2.5 times.
 409                                                 $this->mTables[$variant]->setPair( $from, $to );
 410                                 }
 411                         }
 412                         elseif ( $action == "remove" ) {
 413                                 $this->mTables[$variant]->removeArray( $pair );
 414                         }
 415                 }
 416         }
 417
 418         /**
 419          * Convert text using a parser object for context
 420          * @public
 421          */
 422         function parserConvert( $text, &$parser ) {
 423                 global $wgDisableLangConversion;
 424                 /* don't do anything if this is the conversion table */
 425                 if ( $parser->getTitle()->getNamespace() == NS_MEDIAWIKI &&
 426                                  strpos($parser->mTitle->getText(), "Conversiontable") !== false )
 427                 {
 428                         return $text;
 429                 }
 430
 431                 if ( $wgDisableLangConversion )
 432                         return $text;
 433
 434                 $text = $this->convert( $text );
 435
 436                 if ( $this->mTitleFromFlag )
 437                         $parser->mOutput->setTitleText( $this->mTitleDisplay );
 438                 return $text;
 439         }
 440
 441         /**
 442          * convert namespace
 443          * @param string $title the title included namespace
 444          * @return array of string
 445          * @private
 446          */
 447         function convertNamespace( $title, $variant ) {
 448                 $splittitle = explode( ':', $title );
 449                 if (count($splittitle) < 2)
 450                         return $title;
 451                 if ( isset( $this->mNamespaceTables[$variant][$splittitle[0]] ) )
 452                         $splittitle[0] = $this->mNamespaceTables[$variant][$splittitle[0]];
 453                 $ret = implode(':', $splittitle );
 454                 return $ret;
 455         }
 456
 457         /**
 458          * convert title
 459          * @private
 460          */
 461         function convertTitle( $text, $variant ){
 462                 global $wgDisableTitleConversion, $wgUser;
 463
 464                 // check for global param and __NOTC__ tag
 465                 if( $wgDisableTitleConversion || !$this->mDoTitleConvert || $wgUser->getOption('noconvertlink') == 1 ) {
 466                         $this->mTitleDisplay = $text;
 467                         return $text;
 468                 }
 469
 470                 // use the title from the T flag if any
 471                 if( $this->mTitleFromFlag ){
 472                         $this->mTitleFromFlag = false;
 473                         return $this->mTitleDisplay;
 474                 }
 475
 476                 global $wgRequest;
 477                 $isredir = $wgRequest->getText( 'redirect', 'yes' );
 478                 $action = $wgRequest->getText( 'action' );
 479                 $linkconvert = $wgRequest->getText( 'linkconvert', 'yes' );
 480                 if ( $isredir == 'no' || $action == 'edit' || $action == 'submit' || $linkconvert == 'no' ) {
 481                         return $text;
 482                 } else {
 483                         $text = $this->convertNamespace( $text, $variant );
 484                         $this->mTitleDisplay = $this->convert( $text );
 485                         return $this->mTitleDisplay;
 486                 }
 487         }
 488
 489         /**
 490          * convert text to different variants of a language. the automatic
 491          * conversion is done in autoConvert(). here we parse the text
 492          * marked with -{}-, which specifies special conversions of the
 493          * text that can not be accomplished in autoConvert()
 494          *
 495          * syntax of the markup:
 496          * -{code1:text1;code2:text2;...}-  or
 497          * -{flags|code1:text1;code2:text2;...}-  or
 498          * -{text}- in which case no conversion should take place for text
 499          *
 500          * @param string $text text to be converted
 501          * @param bool $isTitle whether this conversion is for the article title
 502          * @return string converted text
 503          * @public
 504          */
 505         function convert( $text, $isTitle = false ) {
 506
 507                 $mw =& MagicWord::get( 'notitleconvert' );
 508                 if( $mw->matchAndRemove( $text ) )
 509                         $this->mDoTitleConvert = false;
 510                 $mw =& MagicWord::get( 'nocontentconvert' );
 511                 if( $mw->matchAndRemove( $text ) ) {
 512                         $this->mDoContentConvert = false;
 513                 }
 514
 515                 // no conversion if redirecting
 516                 $mw =& MagicWord::get( 'redirect' );
 517                 if( $mw->matchStart( $text ) )
 518                         return $text;
 519
 520                 $plang = $this->getPreferredVariant();
 521
 522                 // for title convertion
 523                 if ( $isTitle ) return $this->convertTitle( $text, $plang );
 524
 525                 $tarray = StringUtils::explode( $this->mMarkup['end'], $text );
 526                 $text = '';
 527
 528                 foreach ( $tarray as $txt ) {
 529
 530                         $marked = explode( $this->mMarkup['begin'], $txt, 2 );
 531
 532                         if( $this->mDoContentConvert )
 533                                 // Bug 19620: should convert a string immediately after a new rule added.
 534                                 $text .= $this->autoConvert( $marked[0], $plang );
 535
 536                         if ( array_key_exists( 1, $marked ) ) {
 537                                 $crule = new ConverterRule($marked[1], $this);
 538                                 $crule->parse( $plang );
 539                                 $text .= $crule->getDisplay();
 540                                 $this->applyManualConv( $crule );
 541                         }
 542                         else
 543                                 $text .= $this->mMarkup['end'];
 544
 545                 }
 546
 547                 // Remove the last delimiter (wasn't real)
 548                 $text = substr( $text, 0, -strlen( $this->mMarkup['end'] ) );
 549
 550                 return $text;
 551         }
 552
 553         /**
 554          * if a language supports multiple variants, it is
 555          * possible that non-existing link in one variant
 556          * actually exists in another variant. this function
 557          * tries to find it. See e.g. LanguageZh.php
 558          *
 559          * @param string $link the name of the link
 560          * @param mixed $nt the title object of the link
 561          * @param boolean $ignoreOtherCond: to disable other conditions when
 562          *      we need to transclude a template or update a category's link
 563          * @return null the input parameters may be modified upon return
 564          * @public
 565          */
 566         function findVariantLink( &$link, &$nt, $ignoreOtherCond = false ) {
 567                 # If the article has already existed, there is no need to
 568                 # check it again, otherwise it may cause a fault.
 569                 if ( $nt->exists() )
 570                         return;
 571
 572                 global $wgDisableLangConversion, $wgDisableTitleConversion, $wgRequest, $wgUser;
 573                 $isredir = $wgRequest->getText( 'redirect', 'yes' );
 574                 $action = $wgRequest->getText( 'action' );
 575                 $linkconvert = $wgRequest->getText( 'linkconvert', 'yes' );
 576                 $disableLinkConversion = $wgDisableLangConversion || $wgDisableTitleConversion;
 577                 $linkBatch = new LinkBatch();
 578
 579                 $ns=NS_MAIN;
 580
 581                 if ( $disableLinkConversion || ( !$ignoreOtherCond && ( $isredir == 'no' || $action == 'edit'
 582                         || $action == 'submit' || $linkconvert == 'no' || $wgUser->getOption('noconvertlink') == 1 ) ) )
 583                         return;
 584
 585                 if ( is_object( $nt ) )
 586                         $ns = $nt->getNamespace();
 587
 588                 $variants = $this->autoConvertToAllVariants($link);
 589                 if($variants == false) //give up
 590                         return;
 591
 592                 $titles = array();
 593
 594                 foreach( $variants as $v ) {
 595                         if($v != $link){
 596                                 $varnt = Title::newFromText( $v, $ns );
 597                                 if(!is_null($varnt)){
 598                                         $linkBatch->addObj($varnt);
 599                                         $titles[]=$varnt;
 600                                 }
 601                         }
 602                 }
 603
 604                 // fetch all variants in single query
 605                 $linkBatch->execute();
 606
 607                 foreach( $titles as $varnt ) {
 608                         if( $varnt->getArticleID() > 0 ) {
 609                                 $nt = $varnt;
 610                                 $link = $varnt->getText();
 611                                 break;
 612                         }
 613                 }
 614         }
 615
 616     /**
 617          * returns language specific hash options
 618          *
 619          * @public
 620          */
 621         function getExtraHashOptions() {
 622                 $variant = $this->getPreferredVariant();
 623                 return '!' . $variant ;
 624         }
 625
 626     /**
 627          * get title text as defined in the body of the article text
 628          *
 629          * @public
 630          */
 631         function getParsedTitle() {
 632                 return $this->mTitleDisplay;
 633         }
 634
 635         /**
 636          * a write lock to the cache
 637          *
 638          * @private
 639          */
 640         function lockCache() {
 641                 global $wgMemc;
 642                 $success = false;
 643                 for($i=0; $i<30; $i++) {
 644                         if($success = $wgMemc->add($this->mCacheKey . "lock", 1, 10))
 645                                 break;
 646                         sleep(1);
 647                 }
 648                 return $success;
 649         }
 650
 651         /**
 652          * unlock cache
 653          *
 654          * @private
 655          */
 656         function unlockCache() {
 657                 global $wgMemc;
 658                 $wgMemc->delete($this->mCacheKey . "lock");
 659         }
 660
 661
 662         /**
 663          * Load default conversion tables
 664          * This method must be implemented in derived class
 665          *
 666          * @private
 667          */
 668         function loadDefaultTables() {
 669                 $name = get_class($this);
 670                 wfDie("Must implement loadDefaultTables() method in class $name");
 671         }
 672
 673         /**
 674          * load conversion tables either from the cache or the disk
 675          * @private
 676          */
 677         function loadTables($fromcache=true) {
 678                 global $wgMemc;
 679                 if( $this->mTablesLoaded )
 680                         return;
 681                 wfProfileIn( __METHOD__ );
 682                 $this->mTablesLoaded = true;
 683                 $this->mTables = false;
 684                 if($fromcache) {
 685                         wfProfileIn( __METHOD__.'-cache' );
 686                         $this->mTables = $wgMemc->get( $this->mCacheKey );
 687                         wfProfileOut( __METHOD__.'-cache' );
 688                 }
 689                 if ( !$this->mTables || !isset( $this->mTables[self::CACHE_VERSION_KEY] ) ) {
 690                         wfProfileIn( __METHOD__.'-recache' );
 691                         // not in cache, or we need a fresh reload.
 692                         // we will first load the default tables
 693                         // then update them using things in MediaWiki:Zhconversiontable/*
 694                         $this->loadDefaultTables();
 695                         foreach($this->mVariants as $var) {
 696                                 $cached = $this->parseCachedTable($var);
 697                                 $this->mTables[$var]->mergeArray($cached);
 698                         }
 699
 700                         $this->postLoadTables();
 701                         $this->mTables[self::CACHE_VERSION_KEY] = true;
 702
 703                         if($this->lockCache()) {
 704                                 $wgMemc->set($this->mCacheKey, $this->mTables, 43200);
 705                                 $this->unlockCache();
 706                         }
 707                         wfProfileOut( __METHOD__.'-recache' );
 708                 }
 709                 wfProfileOut( __METHOD__ );
 710         }
 711
 712     /**
 713          * Hook for post processig after conversion tables are loaded
 714          *
 715          */
 716         function postLoadTables() {}
 717
 718     /**
 719          * Reload the conversion tables
 720          *
 721          * @private
 722          */
 723         function reloadTables() {
 724                 if($this->mTables)
 725                         unset($this->mTables);
 726                 $this->mTablesLoaded = false;
 727                 $this->loadTables(false);
 728         }
 729
 730
 731         /**
 732          * parse the conversion table stored in the cache
 733          *
 734          * the tables should be in blocks of the following form:
 735          *              -{
 736          *                      word => word ;
 737          *                      word => word ;
 738          *                      ...
 739          *              }-
 740          *
 741          *      to make the tables more manageable, subpages are allowed
 742          *      and will be parsed recursively if $recursive=true
 743          *
 744          */
 745         function parseCachedTable($code, $subpage='', $recursive=true) {
 746                 global $wgMessageCache;
 747                 static $parsed = array();
 748
 749                 if(!is_object($wgMessageCache))
 750                         return array();
 751
 752                 $key = 'Conversiontable/'.$code;
 753                 if($subpage)
 754                         $key .= '/' . $subpage;
 755
 756                 if(array_key_exists($key, $parsed))
 757                         return array();
 758
 759                 if ( strpos( $code, '/' ) === false ) {
 760                         $txt = $wgMessageCache->get( 'Conversiontable', true, $code );
 761                 } else {
 762                         $title = Title::makeTitleSafe( NS_MEDIAWIKI, "Conversiontable/$code" );
 763                         if ( $title && $title->exists() ) {
 764                                 $article = new Article( $title );
 765                                 $txt = $article->getContents();
 766                         } else {
 767                                 $txt = '';
 768                         }
 769                 }
 770
 771                 // get all subpage links of the form
 772                 // [[MediaWiki:conversiontable/zh-xx/...|...]]
 773                 $linkhead = $this->mLangObj->getNsText(NS_MEDIAWIKI) . ':Conversiontable';
 774                 $subs = explode('[[', $txt);
 775                 $sublinks = array();
 776                 foreach( $subs as $sub ) {
 777                         $link = explode(']]', $sub, 2);
 778                         if(count($link) != 2)
 779                                 continue;
 780                         $b = explode('|', $link[0]);
 781                         $b = explode('/', trim($b[0]), 3);
 782                         if(count($b)==3)
 783                                 $sublink = $b[2];
 784                         else
 785                                 $sublink = '';
 786
 787                         if($b[0] == $linkhead && $b[1] == $code) {
 788                                 $sublinks[] = $sublink;
 789                         }
 790                 }
 791
 792
 793                 // parse the mappings in this page
 794                 $blocks = explode($this->mMarkup['begin'], $txt);
 795                 array_shift($blocks);
 796                 $ret = array();
 797                 foreach($blocks as $block) {
 798                         $mappings = explode($this->mMarkup['end'], $block, 2);
 799                         $stripped = str_replace(array("'", '"', '*','#'), '', $mappings[0]);
 800                         $table = explode( ';', $stripped );
 801                         foreach( $table as $t ) {
 802                                 $m = explode( '=>', $t );
 803                                 if( count( $m ) != 2)
 804                                         continue;
 805                                 // trim any trailling comments starting with '//'
 806                                 $tt = explode('//', $m[1], 2);
 807                                 $ret[trim($m[0])] = trim($tt[0]);
 808                         }
 809                 }
 810                 $parsed[$key] = true;
 811
 812
 813                 // recursively parse the subpages
 814                 if($recursive) {
 815                         foreach($sublinks as $link) {
 816                                 $s = $this->parseCachedTable($code, $link, $recursive);
 817                                 $ret = array_merge($ret, $s);
 818                         }
 819                 }
 820
 821                 if ($this->mUcfirst) {
 822                         foreach ($ret as $k => $v) {
 823                                 $ret[Language::ucfirst($k)] = Language::ucfirst($v);
 824                         }
 825                 }
 826                 return $ret;
 827         }
 828
 829         /**
 830          * Enclose a string with the "no conversion" tag. This is used by
 831          * various functions in the Parser
 832          *
 833          * @param string $text text to be tagged for no conversion
 834          * @return string the tagged text
 835          * @public
 836          */
 837         function markNoConversion($text, $noParse=false) {
 838                 # don't mark if already marked
 839                 if(strpos($text, $this->mMarkup['begin']) ||
 840                    strpos($text, $this->mMarkup['end']))
 841                         return $text;
 842
 843                 $ret = $this->mMarkup['begin'] .'R|'. $text . $this->mMarkup['end'];
 844                 return $ret;
 845         }
 846
 847         /**
 848          * convert the sorting key for category links. this should make different
 849          * keys that are variants of each other map to the same key
 850          */
 851         function convertCategoryKey( $key ) {
 852                 return $key;
 853         }
 854         /**
 855          * hook to refresh the cache of conversion tables when
 856          * MediaWiki:conversiontable* is updated
 857          * @private
 858          */
 859         function OnArticleSaveComplete($article, $user, $text, $summary, $isminor, $iswatch, $section, $flags, $revision) {
 860                 $titleobj = $article->getTitle();
 861                 if($titleobj->getNamespace() == NS_MEDIAWIKI) {
 862                         $title = $titleobj->getDBkey();
 863                         $t = explode('/', $title, 3);
 864                         $c = count($t);
 865                         if( $c > 1 && $t[0] == 'Conversiontable' ) {
 866                                 if(in_array($t[1], $this->mVariants)) {
 867                                         $this->reloadTables();
 868                                 }
 869                         }
 870                 }
 871                 return true;
 872         }
 873
 874         /**
 875          * Armour rendered math against conversion
 876          * Wrap math into rawoutput -{R| math }- syntax
 877          * @public
 878          */
 879         function armourMath($text){
 880                 // we need to convert '-{' and '}-' to '-&#123;' and '&#125;-'
 881                 // to avoid a unwanted '}-' appeared after the math-image.
 882                 $text = strtr( $text, array('-{' => '-&#123;', '}-' => '&#125;-') );
 883                 $ret = $this->mMarkup['begin'] . 'R|' . $text . $this->mMarkup['end'];
 884                 return $ret;
 885         }
 886 }
 887
 888 /**
 889  * parser for rules of language conversion , parse rules in -{ }- tag
 890  * @ingroup Language
 891  * @author fdcn <fdcn64@gmail.com>, PhiLiP <philip.npc@gmail.com>
 892  */
 893 class ConverterRule {
 894         var $mText; // original text in -{text}-
 895         var $mConverter; // LanguageConverter object
 896         var $mManualCodeError='<strong class="error">code error!</strong>';
 897         var $mRuleDisplay = '',$mRuleTitle=false;
 898         var $mRules = '';// string : the text of the rules
 899         var $mRulesAction = 'none';
 900         var $mFlags = array();
 901         var $mConvTable = array();
 902         var $mBidtable = array();// array of the translation in each variant
 903         var $mUnidtable = array();// array of the translation in each variant
 904
 905         /**
 906          * Constructor
 907          *
 908          * @param string $text the text between -{ and }-
 909          * @param object $converter a  LanguageConverter object
 910          * @access public
 911          */
 912         function __construct($text,$converter){
 913                 $this->mText = $text;
 914                 $this->mConverter=$converter;
 915                 foreach($converter->mVariants as $v){
 916                         $this->mConvTable[$v]=array();
 917                 }
 918         }
 919
 920         /**
 921          * check if variants array in convert array
 922          *
 923          * @param string $variant Variant language code
 924          * @return string Translated text
 925          * @public
 926          */
 927         function getTextInBidtable($variants){
 928                 if(is_string($variants)){ $variants=array($variants); }
 929                 if(!is_array($variants)) return false;
 930                 foreach ($variants as $variant){
 931                         if(array_key_exists($variant, $this->mBidtable)){
 932                                 return $this->mBidtable[$variant];
 933                         }
 934                 }
 935                 return false;
 936         }
 937
 938         /**
 939          * Parse flags with syntax -{FLAG| ... }-
 940          * @private
 941          */
 942         function parseFlags(){
 943                 $text = $this->mText;
 944                 if(strlen($text) < 2 ) {
 945                         $this->mFlags = array( 'R' );
 946                         $this->mRules = $text;
 947                         return;
 948                 }
 949
 950                 $flags = array();
 951                 $markup = $this->mConverter->mMarkup;
 952                 $validFlags = $this->mConverter->mFlags;
 953                 $variants = $this->mConverter->mVariants;
 954
 955                 $tt = explode($markup['flagsep'], $text, 2);
 956                 if(count($tt) == 2) {
 957                         $f = explode($markup['varsep'], $tt[0]);
 958                         foreach($f as $ff) {
 959                                 $ff = trim($ff);
 960                                 if(array_key_exists($ff, $validFlags) &&
 961                                                         !in_array($validFlags[$ff], $flags))
 962                                         $flags[] = $validFlags[$ff];
 963                         }
 964                         $rules = $tt[1];
 965                 } else {
 966                         $rules = $text;
 967                 }
 968
 969                 //check flags
 970                 if( in_array('R',$flags) ){
 971                         $flags = array('R');// remove other flags
 972                 } elseif ( in_array('N',$flags) ){
 973                         $flags = array('N');// remove other flags
 974                 } elseif ( in_array('-',$flags) ){
 975                         $flags = array('-');// remove other flags
 976                 } elseif (count($flags)==1 && $flags[0]=='T'){
 977                         $flags[]='H';
 978                 } elseif ( in_array('H',$flags) ){
 979                         // replace A flag, and remove other flags except T
 980                         $temp=array('+','H');
 981                         if(in_array('T',$flags)) $temp[] = 'T';
 982                         if(in_array('D',$flags)) $temp[] = 'D';
 983                         $flags = $temp;
 984                 } else {
 985                         if ( in_array('A',$flags) ) {
 986                                 $flags[]='+';
 987                                 $flags[]='S';
 988                         }
 989                         if ( in_array('D',$flags) )
 990                                 $flags=array_diff($flags,array('S'));
 991                         $flags_temp = array();
 992                         foreach ($variants as $variant) {
 993                                 // try to find flags like "zh-hans", "zh-hant"
 994                                 // allow syntaxes like "-{zh-hans;zh-hant|XXXX}-"
 995                                 if ( in_array($variant, $flags) )
 996                                         $flags_temp[] = $variant;
 997                         }
 998                         if ( count($flags_temp) !== 0 )
 999                                 $flags = $flags_temp;
1000                 }
1001                 if ( count($flags) == 0 )
1002                         $flags = array('S');
1003                 $this->mRules=$rules;
1004                 $this->mFlags=$flags;
1005         }
1006
1007         /**
1008          * generate conversion table
1009          * @private
1010          */
1011         function parseRules() {
1012                 $rules = $this->mRules;
1013                 $flags = $this->mFlags;
1014                 $bidtable = array();
1015                 $unidtable = array();
1016                 $markup = $this->mConverter->mMarkup;
1017                 $variants = $this->mConverter->mVariants;
1018
1019                 // varsep_pattern for preg_split:
1020                 // text should be splited by ";" only if a valid variant
1021                 // name exist after the markup, for example:
1022                 //  -{zh-hans:<span style="font-size:120%;">xxx</span>;zh-hant:<span style="font-size:120%;">yyy</span>;}-
1023                 // we should split it as:
1024                 //  array(
1025                 //        [0] => 'zh-hans:<span style="font-size:120%;">xxx</span>'
1026                 //        [1] => 'zh-hant:<span style="font-size:120%;">yyy</span>'
1027                 //        [2] => ''
1028                 //       )
1029                 $varsep_pattern = '/' . $markup['varsep'] . '\s*' . '(?=';
1030                 foreach( $variants as $variant ) {
1031                         $varsep_pattern .= $variant . '\s*' . $markup['codesep'] . '|'; // zh-hans:xxx;zh-hant:yyy
1032                         $varsep_pattern .= '[^;]*?' . $markup['unidsep'] . '\s*' . $variant
1033                                                         . '\s*' . $markup['codesep'] . '|'; // xxx=>zh-hans:yyy; xxx=>zh-hant:zzz
1034                 }
1035                 $varsep_pattern .= '\s*$)/';
1036
1037                 $choice = preg_split($varsep_pattern, $rules);
1038
1039                 foreach( $choice as $c ) {
1040                         $v  = explode($markup['codesep'], $c, 2);
1041                         if( count($v) != 2 )
1042                                 continue;// syntax error, skip
1043                         $to = trim($v[1]);
1044                         $v  = trim($v[0]);
1045                         $u  = explode($markup['unidsep'], $v);
1046                         // if $to is empty, strtr() could return a wrong result
1047                         if( count($u) == 1 && $to && in_array( $v, $variants ) ) {
1048                                 $bidtable[$v] = $to;
1049                         } else if(count($u) == 2){
1050                                 $from = trim($u[0]);
1051                                 $v    = trim($u[1]);
1052                                 if( array_key_exists( $v, $unidtable ) && !is_array( $unidtable[$v] )
1053                                         && $to && in_array( $v, $variants ) )
1054                                         $unidtable[$v] = array( $from=>$to );
1055                                 elseif ( $to && in_array( $v, $variants ) )
1056                                         $unidtable[$v][$from] = $to;
1057                         }
1058                         // syntax error, pass
1059                         if ( !array_key_exists( $v, $this->mConverter->mVariantNames ) ){
1060                                 $bidtable = array();
1061                                 $unidtable = array();
1062                                 break;
1063                         }
1064                 }
1065                 $this->mBidtable = $bidtable;
1066                 $this->mUnidtable = $unidtable;
1067         }
1068
1069         /**
1070          * @private
1071          */
1072         function getRulesDesc(){
1073                 $codesep = $this->mConverter->mDescCodeSep;
1074                 $varsep = $this->mConverter->mDescVarSep;
1075                 $text='';
1076                 foreach($this->mBidtable as $k => $v)
1077                         $text .= $this->mConverter->mVariantNames[$k]."$codesep$v$varsep";
1078                 foreach($this->mUnidtable as $k => $a)
1079                         foreach($a as $from=>$to)
1080                                 $text.=$from.'⇒'.$this->mConverter->mVariantNames[$k]."$codesep$to$varsep";
1081                 return $text;
1082         }
1083
1084         /**
1085          *  Parse rules conversion
1086          * @private
1087          */
1088         function getRuleConvertedStr($variant,$doConvert){
1089                 $bidtable = $this->mBidtable;
1090                 $unidtable = $this->mUnidtable;
1091
1092                 if( count($bidtable) + count($unidtable) == 0 ){
1093                         return $this->mRules;
1094                 } elseif ($doConvert){// the text converted
1095                         // display current variant in bidirectional array
1096                         $disp = $this->getTextInBidtable($variant);
1097                         // or display current variant in fallbacks
1098                         if(!$disp)
1099                                 $disp = $this->getTextInBidtable(
1100                                                 $this->mConverter->getVariantFallbacks($variant));
1101                         // or display current variant in unidirectional array
1102                         if(!$disp && array_key_exists($variant,$unidtable)){
1103                                 $disp = array_values($unidtable[$variant]);
1104                                 $disp = $disp[0];
1105                         }
1106                         // or display frist text under disable manual convert
1107                         if(!$disp && $this->mConverter->mManualLevel[$variant]=='disable') {
1108                                 if(count($bidtable)>0){
1109                                         $disp = array_values($bidtable);
1110                                         $disp = $disp[0];
1111                                 } else {
1112                                         $disp = array_values($unidtable);
1113                                         $disp = array_values($disp[0]);
1114                                         $disp = $disp[0];
1115                                 }
1116                         }
1117                         return $disp;
1118                 } else {// no convert
1119                         return $this->mRules;
1120                 }
1121         }
1122
1123         /**
1124          * generate conversion table for all text
1125          * @private
1126          */
1127         function generateConvTable(){
1128                 $flags = $this->mFlags;
1129                 $bidtable = $this->mBidtable;
1130                 $unidtable = $this->mUnidtable;
1131                 $manLevel = $this->mConverter->mManualLevel;
1132
1133                 $vmarked=array();
1134                 foreach($this->mConverter->mVariants as $v) {
1135                         /* for bidirectional array
1136                                 fill in the missing variants, if any,
1137                                 with fallbacks */
1138                         if(!array_key_exists($v, $bidtable)) {
1139                                 $variantFallbacks = $this->mConverter->getVariantFallbacks($v);
1140                                 $vf = $this->getTextInBidtable($variantFallbacks);
1141                                 if($vf) $bidtable[$v] = $vf;
1142                         }
1143
1144                         if(array_key_exists($v,$bidtable)){
1145                                 foreach($vmarked as $vo){
1146                                         // use syntax: -{A|zh:WordZh;zh-tw:WordTw}-
1147                                         // or -{H|zh:WordZh;zh-tw:WordTw}- or -{-|zh:WordZh;zh-tw:WordTw}-
1148                                         // to introduce a custom mapping between
1149                                         // words WordZh and WordTw in the whole text
1150                                         if($manLevel[$v]=='bidirectional'){
1151                                                 $this->mConvTable[$v][$bidtable[$vo]]=$bidtable[$v];
1152                                         }
1153                                         if($manLevel[$vo]=='bidirectional'){
1154                                                 $this->mConvTable[$vo][$bidtable[$v]]=$bidtable[$vo];
1155                                         }
1156                                 }
1157                                 $vmarked[]=$v;
1158                         }
1159                         /*for unidirectional array
1160                                 fill to convert tables */
1161                         $allow_unid = $manLevel[$v]=='bidirectional'
1162                                         || $manLevel[$v]=='unidirectional';
1163                         if($allow_unid && array_key_exists($v,$unidtable)){
1164                                 $ct=$this->mConvTable[$v];
1165                                 $this->mConvTable[$v] = array_merge($ct,$unidtable[$v]);
1166                         }
1167                 }
1168         }
1169
1170         /**
1171          * Parse rules and flags
1172          * @public
1173          */
1174         function parse($variant){
1175                 if(!$variant)
1176                         $variant = $this->mConverter->getPreferredVariant();
1177
1178                 $variants = $this->mConverter->mVariants;
1179                 $this->parseFlags();
1180                 $flags = $this->mFlags;
1181
1182                 // convert to specified variant
1183                 // syntax: -{zh-hans;zh-hant[;...]|<text to convert>}-
1184                 if( count( array_diff( $flags, $variants ) ) == 0 and count( $flags ) != 0 ) {
1185                         if ( in_array( $variant, $flags ) ) // check if current variant in flags
1186                                 // then convert <text to convert> to current language
1187                                 $this->mRules = $this->mConverter->autoConvert( $this->mRules, $variant );
1188                         else { // if current variant no in flags,
1189                                    // then we check its fallback variants.
1190                                 $variantFallbacks = $this->mConverter->getVariantFallbacks($variant);
1191                                 foreach ( $variantFallbacks as $variantFallback ) {
1192                                         // if current variant's fallback exist in flags
1193                                         if ( in_array( $variantFallback, $flags ) ) {
1194                                                 // then convert <text to convert> to fallback language
1195                                                 $this->mRules = $this->mConverter->autoConvert( $this->mRules, $variantFallback );
1196                                                 break;
1197                                         }
1198                                 }
1199                         }
1200                         $this->mFlags = $flags = array('R');
1201                 }
1202
1203                 if( !in_array( 'R', $flags ) || !in_array( 'N', $flags ) ) {
1204                         // decode => HTML entities modified by Sanitizer::removeHTMLtags
1205                         $this->mRules = str_replace('=&gt;','=>',$this->mRules);
1206
1207                         $this->parseRules();
1208                 }
1209                 $rules = $this->mRules;
1210
1211                 if( count( $this->mBidtable ) == 0 && count( $this->mUnidtable ) == 0 ){
1212                         if(in_array('+',$flags) || in_array('-',$flags))
1213                                 // fill all variants if text in -{A/H/-|text} without rules
1214                                 foreach($this->mConverter->mVariants as $v)
1215                                         $this->mBidtable[$v] = $rules;
1216                         elseif (!in_array('N',$flags) && !in_array('T',$flags) )
1217                                 $this->mFlags = $flags = array('R');
1218                 }
1219
1220                 if( in_array('R',$flags) ) {
1221                         // if we don't do content convert, still strip the -{}- tags
1222                         $this->mRuleDisplay = $rules;
1223                 } elseif ( in_array('N',$flags) ){
1224                         // proces N flag: output current variant name
1225                         $this->mRuleDisplay = $this->mConverter->mVariantNames[trim($rules)];
1226                 } elseif ( in_array('D',$flags) ){
1227                         // proces D flag: output rules description
1228                         $this->mRuleDisplay = $this->getRulesDesc();
1229                 } elseif ( in_array('H',$flags) || in_array('-',$flags) ) {
1230                         // proces H,- flag or T only: output nothing
1231                         $this->mRuleDisplay = '';
1232                 } elseif ( in_array('S',$flags) ){
1233                         $this->mRuleDisplay = $this->getRuleConvertedStr($variant,
1234                                                         $this->mConverter->mDoContentConvert);
1235                 } else {
1236                         $this->mRuleDisplay= $this->mManualCodeError;
1237                 }
1238                 // proces T flag
1239                 if ( in_array('T',$flags) ) {
1240                         $this->mRuleTitle = $this->getRuleConvertedStr($variant,
1241                                                         $this->mConverter->mDoTitleConvert);
1242                 }
1243
1244                 if (in_array('-', $flags))
1245                         $this->mRulesAction='remove';
1246                 if (in_array('+', $flags))
1247                         $this->mRulesAction='add';
1248
1249                 $this->generateConvTable();
1250         }
1251
1252         /**
1253          * @public
1254          */
1255         function hasRules(){
1256                 // TODO:
1257         }
1258
1259         /**
1260          * get display text on markup -{...}-
1261          * @public
1262          */
1263         function getDisplay(){
1264                 return $this->mRuleDisplay;
1265         }
1266         /**
1267          * get converted title
1268          * @public
1269          */
1270         function getTitle(){
1271                 return $this->mRuleTitle;
1272         }
1273
1274         /**
1275          * return how deal with conversion rules
1276          * @public
1277          */
1278         function getRulesAction(){
1279                 return $this->mRulesAction;
1280         }
1281
1282         /**
1283          * get conversion table ( bidirectional and unidirectional conversion table )
1284          * @public
1285          */
1286         function getConvTable(){
1287                 return $this->mConvTable;
1288         }
1289
1290         /**
1291          * get conversion rules string
1292          * @public
1293          */
1294         function getRules(){
1295                 return $this->mRules;
1296         }
1297
1298         /**
1299          * get conversion flags
1300          * @public
1301          */
1302         function getFlags(){
1303                 return $this->mFlags;
1304         }
1305 }