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