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