languages/LanguageConverter.php

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