languages/LanguageConverter.php

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