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