languages/LanguageConverter.php

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