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                 // disable conversion of <pre xxxx> ... </pre>
 235                 $prefix = '<pre.*?>.*?<\/pre>|';
 236
 237                 $reg = '/'.$codefix . $scriptfix . $prefix . '<[^>]+>|&[a-zA-Z#][a-z0-9]+;' . $marker . $htmlfix . '/s';
 238
 239                 $matches = preg_split($reg, $text, -1, PREG_SPLIT_OFFSET_CAPTURE);
 240
 241                 $m = array_shift($matches);
 242
 243                 $ret = $this->translate($m[0], $toVariant);
 244                 $mstart = $m[1]+strlen($m[0]);
 245                 foreach($matches as $m) {
 246                         $ret .= substr($text, $mstart, $m[1]-$mstart);
 247                         $ret .= $this->translate($m[0], $toVariant);
 248                         $mstart = $m[1] + strlen($m[0]);
 249                 }
 250                 wfProfileOut( $fname );
 251                 return $ret;
 252         }
 253
 254         /**
 255          * Translate a string to a variant
 256          * Doesn't process markup or do any of that other stuff, for that use convert()
 257          *
 258          * @param string $text Text to convert
 259          * @param string $variant Variant language code
 260          * @return string Translated text
 261          * @private
 262          */
 263         function translate( $text, $variant ) {
 264                 wfProfileIn( __METHOD__ );
 265                 if( !$this->mTablesLoaded )
 266                         $this->loadTables();
 267                 $text = $this->mTables[$variant]->replace( $text );
 268                 wfProfileOut( __METHOD__ );
 269                 return $text;
 270         }
 271
 272         /**
 273          * convert text to all supported variants
 274          *
 275          * @param string $text the text to be converted
 276          * @return array of string
 277          * @public
 278          */
 279         function autoConvertToAllVariants($text) {
 280                 $fname="LanguageConverter::autoConvertToAllVariants";
 281                 wfProfileIn( $fname );
 282                 if( !$this->mTablesLoaded )
 283                         $this->loadTables();
 284
 285                 $ret = array();
 286                 foreach($this->mVariants as $variant) {
 287                         $ret[$variant] = $this->translate($text, $variant);
 288                 }
 289
 290                 wfProfileOut( $fname );
 291                 return $ret;
 292         }
 293
 294         /**
 295          * convert link text to all supported variants
 296          *
 297          * @param string $text the text to be converted
 298          * @return array of string
 299          * @public
 300          */
 301         function convertLinkToAllVariants($text) {
 302                 if( !$this->mTablesLoaded )
 303                         $this->loadTables();
 304
 305                 $ret = array();
 306                 $tarray = explode($this->mMarkup['begin'], $text);
 307                 $tfirst = array_shift($tarray);
 308
 309                 foreach($this->mVariants as $variant)
 310                         $ret[$variant] = $this->translate($tfirst,$variant);
 311
 312                 foreach($tarray as $txt) {
 313                         $marked = explode($this->mMarkup['end'], $txt, 2);
 314
 315                         foreach($this->mVariants as $variant){
 316                                 $ret[$variant] .= $this->mMarkup['begin'].$marked[0].$this->mMarkup['end'];
 317                                 if(array_key_exists(1, $marked))
 318                                         $ret[$variant] .= $this->translate($marked[1],$variant);
 319                         }
 320
 321                 }
 322
 323                 return $ret;
 324         }
 325
 326
 327         /**
 328          * Convert text using a parser object for context
 329          * @public
 330          */
 331         function parserConvert( $text, &$parser ) {
 332                 global $wgDisableLangConversion;
 333                 /* don't do anything if this is the conversion table */
 334                 if ( $parser->getTitle()->getNamespace() == NS_MEDIAWIKI &&
 335                                  strpos($parser->mTitle->getText(), "Conversiontable") !== false )
 336                 {
 337                         return $text;
 338                 }
 339
 340                 if($wgDisableLangConversion)
 341                         return $text;
 342
 343                 $text = $this->convert( $text );
 344                 $parser->mOutput->setTitleText( $this->mTitleDisplay );
 345                 return $text;
 346         }
 347
 348         /**
 349          *  Parse flags with syntax -{FLAG| ... }-
 350          * @private
 351          */
 352         function parseFlags($marked){
 353                 $flags = array();
 354
 355                 // for multi-FLAGs
 356                 if(strlen($marked) < 2 )
 357                         return array($marked,array('R'));
 358
 359                 $tt = explode($this->mMarkup['flagsep'], $marked, 2);
 360
 361                 if(count($tt) == 2) {
 362                         $f = explode($this->mMarkup['varsep'], $tt[0]);
 363                         foreach($f as $ff) {
 364                                 $ff = trim($ff);
 365                                 if(array_key_exists($ff, $this->mFlags) &&
 366                                                         !in_array($this->mFlags[$ff], $flags))
 367                                         $flags[] = $this->mFlags[$ff];
 368                         }
 369                         $rules = $tt[1];
 370                 } else {
 371                         $rules = $marked;
 372                 }
 373
 374                 if( !in_array('R',$flags) ){
 375                         //FIXME: may cause trouble here...
 376                         //strip &nbsp; since it interferes with the parsing, plus,
 377                         //all spaces should be stripped in this tag anyway.
 378                         $rules = str_replace('&nbsp;', '', $rules);
 379                         $rules = str_replace('=&gt;','=>',$rules);
 380                 }
 381
 382                 //check flags
 383                 if( in_array('R',$flags) ){
 384                         $flags = array('R');// remove other flags
 385                 } elseif ( in_array('N',$flags) ){
 386                         $flags = array('N');// remove other flags
 387                 } elseif ( in_array('-',$flags) ){
 388                         $flags = array('-');// remove other flags
 389                 } elseif (count($flags)==1 && $flags[0]=='T'){
 390                         $flags[]='H';
 391                 } elseif ( in_array('H',$flags) ){
 392                         // replace A flag, and remove other flags except T
 393                         $temp=array('+','H');
 394                         if(in_array('T',$flags)) $temp[] = 'T';
 395                         if(in_array('D',$flags)) $temp[] = 'D';
 396                         $flags = $temp;
 397                 } else {
 398                         if ( in_array('A',$flags)) {
 399                                 $flags[]='+';
 400                                 $flags[]='S';
 401                         }
 402                         if ( in_array('D',$flags) )
 403                                 $flags=array_diff($flags,array('S'));
 404                 }
 405                 if ( count($flags)==0 )
 406                         $flags = array('S');
 407                 return array($rules,$flags);
 408         }
 409
 410         /**
 411          * @private
 412          */
 413         function getRulesDesc($bidtable,$unidtable){
 414                 $text='';
 415                 foreach($bidtable as $k => $v)
 416                         $text .= $this->mVariantNames[$k].':'.$v.';';
 417                 foreach($unidtable as $k => $a)
 418                         foreach($a as $from=>$to)
 419                                 $text.=$from.'⇒'.$this->mVariantNames[$k].':'.$to.';';
 420                 return $text;
 421         }
 422
 423         /**
 424          * parse the manually marked conversion rule
 425          * @param string $rule the text of the rule
 426          * @return array of the translation in each variant
 427          * @private
 428          */
 429         function getConvTableFromRules($rules,$flags=array()) {
 430                 $bidtable = array();
 431                 $unidtable = array();
 432                 $choice = explode($this->mMarkup['varsep'], $rules );
 433                 foreach($choice as $c) {
 434                         $v = explode($this->mMarkup['codesep'], $c);
 435                         if(count($v) != 2)
 436                                 continue;// syntax error, skip
 437                         $to=trim($v[1]);
 438                         $v=trim($v[0]);
 439                         $u = explode($this->mMarkup['unidsep'], $v);
 440                         if(count($u) == 1) {
 441                                 $bidtable[$v] = $to;
 442                         } else if(count($u) == 2){
 443                                 $from=trim($u[0]);$v=trim($u[1]);
 444                                 if( array_key_exists($v,$unidtable) && !is_array($unidtable[$v]) )
 445                                         $unidtable[$v]=array($from=>$to);
 446                                 else
 447                                         $unidtable[$v][$from]=$to;
 448                         }
 449                         // syntax error, pass
 450                 }
 451                 return array($bidtable,$unidtable);
 452         }
 453
 454         /**
 455          *  get display text on markup -{...}-
 456          * @param string $rules the original code
 457          * @param array $flags FLAGs
 458          * @param array $bidtable bidirectional convert table
 459          * @param string $unidtable unidirectional convert table
 460          * @param string $variant the current variant
 461          * @param bool $$doConvert if do convert
 462          * @private
 463          */
 464         function getRulesDisplay($rules,$flags,
 465                                                         $bidtable,$unidtable,
 466                                                         $variant=false,$doConvert=true){
 467                 if(!$variant) $variant = $this->getPreferredVariant();
 468                 $is_mc_disable = $this->mManualLevel[$variant]=='disable';
 469
 470                 if( in_array('R',$flags) ) {
 471                         // if we don't do content convert, still strip the -{}- tags
 472                         $disp = $rules;
 473                 } elseif ( in_array('N',$flags) ){
 474                         // proces N flag: output current variant name
 475                         $disp = $this->mVariantNames[trim($rules)];
 476                 } elseif ( in_array('D',$flags) ){
 477                         // proces D flag: output rules description
 478                         $disp = $this->getRulesDesc($bidtable,$unidtable);
 479                 } elseif ( in_array('H',$flags) || in_array('-',$flags) ) {
 480                         // proces H,- flag or T only: output nothing
 481                         $disp = '';
 482                 } elseif ( in_array('S',$flags) ){
 483                         if( count($bidtable) + count($unidtable) == 0 ){
 484                                 $disp = $rules;
 485                         } elseif ($doConvert){// the text converted
 486                                 // display current variant in bidirectional array
 487                                 $disp = $this->getTextInCArray($variant,$bidtable);
 488                                 // or display current variant in fallbacks
 489                                 if(!$disp)
 490                                         $disp = $this->getTextInCArray($this->getVariantFallbacks($variant),$bidtable);
 491                                 // or display current variant in unidirectional array
 492                                 if(!$disp && array_key_exists($variant,$unidtable)){
 493                                         $disp = array_values($unidtable[$variant]);
 494                                         $disp = $disp[0];
 495                                 }
 496                                 // or display frist text under disable manual convert
 497                                 if(!$disp && $is_mc_disable) {
 498                                         if(count($bidtable)>0){
 499                                                 $disp = array_values($bidtable);
 500                                                 $disp = $disp[0];
 501                                         } else {
 502                                                 $disp = array_values($unidtable);
 503                                                 $disp = array_values($disp[0]);
 504                                                 $disp = $disp[0];
 505                                         }
 506                                 }
 507                         } else {// no convert
 508                                 $disp = $rules;
 509                         }
 510                 } elseif ( in_array('T',$flags) ) {
 511                         // proces T flag : output nothing
 512                         $disp = '';
 513                 }
 514                 else
 515                         $disp= $this->mManualCodeError;
 516
 517                 return $disp;
 518         }
 519
 520         /**
 521          * @access private
 522          */
 523         function applyManualFlag($rules,$flags,$bidtable,$unidtable,$variant=false){
 524                 if(!$variant) $variant = $this->getPreferredVariant();
 525
 526                 $is_title_flag = in_array('T',  $flags);
 527                 // use syntax -{T|zh:TitleZh;zh-tw:TitleTw}- for custom conversion in title
 528                 if($is_title_flag){
 529                         $this->mTitleFromFlag = true;
 530                         $this->mTitleDisplay =  $this->getRulesDisplay($rules,array('S'),
 531                                                                                                 $bidtable,$unidtable,
 532                                                                                                 $variant,
 533                                                                                                 $this->mDoTitleConvert);
 534                 }
 535
 536                 if($this->mManualLevel[$variant]=='disable') return;
 537
 538                 $is_remove_flag = !$is_title_flag && in_array('-', $flags);
 539                 $is_add_flag = !$is_remove_flag && in_array('+', $flags);
 540                 $is_bidMC = $this->mManualLevel[$variant]=='bidirectional';
 541                 $is_unidMC = $this->mManualLevel[$variant]=='unidirectional';
 542                 $vmarked=array();
 543
 544                 foreach($this->mVariants as $v) {
 545                         /* for bidirectional array
 546                                 fill in the missing variants, if any,
 547                                 with fallbacks */
 548                         if($is_bidMC && !array_key_exists($v, $bidtable)) {
 549                                 $vf = $this->getTextInCArray($this->getVariantFallbacks($v),$bidtable);
 550                                 if($vf) $bidtable[$v] = $vf;
 551                         }
 552                         if($is_bidMC && array_key_exists($v,$bidtable)){
 553                                 foreach($vmarked as $vo){
 554                                         // use syntax:
 555                                         //  -{A|zh:WordZh;zh-tw:WordTw}- or -{+|zh:WordZh;zh-tw:WordTw}-
 556                                         // to introduce a custom mapping between
 557                                         // words WordZh and WordTw in the whole text
 558                                         if($is_add_flag){
 559                                                 $this->mTables[$v]->setPair($bidtable[$vo], $bidtable[$v]);
 560                                                 $this->mTables[$vo]->setPair($bidtable[$v], $bidtable[$vo]);
 561                                         }
 562                                         // use syntax -{-|zh:WordZh;zh-tw:WordTw}- to remove a conversion
 563                                         // words WordZh and WordTw in the whole text
 564                                         if($is_remove_flag){
 565                                                 $this->mTables[$v]->removePair($bidtable[$vo]);
 566                                                 $this->mTables[$vo]->removePair($bidtable[$v]);
 567                                         }
 568                                 }
 569                                 $vmarked[]=$v;
 570                         }
 571                         /*for unidirectional array
 572                                 fill to convert tables */
 573                         if($is_unidMC && array_key_exists($v,$unidtable)){
 574                                 if($is_add_flag)$this->mTables[$v]->mergeArray($unidtable[$v]);
 575                                 if($is_remove_flag)$this->mTables[$v]->removeArray($unidtable[$v]);
 576                         }
 577                 }
 578         }
 579
 580         /**
 581          *  Parse rules and flags
 582          * @private
 583          */
 584         function parseRules($rules,$flags,$variant=false){
 585                 if(!$variant) $variant = $this->getPreferredVariant();
 586
 587                 list($bidtable,$unidtable) = $this->getConvTableFromRules($rules, $flags);
 588                 if(count($bidtable)==0 && count($unidtable)==0
 589                         && !in_array('N',$flags) && !in_array('T',$flags) )
 590                                 $flags = array('R');
 591                 $disp = $this->getRulesDisplay($rules,$flags,
 592                                                                         $bidtable,$unidtable,
 593                                                                         $variant,
 594                                                                         $this->mDoContentConvert);
 595                 $this->applyManualFlag($rules,$flags,$bidtable,$unidtable);
 596
 597                 return $disp;
 598         }
 599
 600         /**
 601          *  convert title
 602          * @private
 603          */
 604         function convertTitle($text){
 605                 // check for __NOTC__ tag
 606                 if( !$this->mDoTitleConvert ) {
 607                         $this->mTitleDisplay = $text;
 608                         return $text;
 609                 }
 610
 611                 // use the title from the T flag if any
 612                 if($this->mTitleFromFlag){
 613                         $this->mTitleFromFlag = false;
 614                         return $this->mTitleDisplay;
 615                 }
 616
 617                 global $wgRequest;
 618                 $isredir = $wgRequest->getText( 'redirect', 'yes' );
 619                 $action = $wgRequest->getText( 'action' );
 620                 if ( $isredir == 'no' || $action == 'edit' ) {
 621                         return $text;
 622                 } else {
 623                         $this->mTitleDisplay = $this->convert($text);
 624                         return $this->mTitleDisplay;
 625                 }
 626         }
 627
 628         /**
 629          * convert text to different variants of a language. the automatic
 630          * conversion is done in autoConvert(). here we parse the text
 631          * marked with -{}-, which specifies special conversions of the
 632          * text that can not be accomplished in autoConvert()
 633          *
 634          * syntax of the markup:
 635          * -{code1:text1;code2:text2;...}-  or
 636          * -{flags|code1:text1;code2:text2;...}-  or
 637          * -{text}- in which case no conversion should take place for text
 638          *
 639          * @param string $text text to be converted
 640          * @param bool $isTitle whether this conversion is for the article title
 641          * @return string converted text
 642          * @access public
 643          */
 644         function convert( $text , $isTitle=false) {
 645
 646                 $mw =& MagicWord::get( 'notitleconvert' );
 647                 if( $mw->matchAndRemove( $text ) )
 648                         $this->mDoTitleConvert = false;
 649                 $mw =& MagicWord::get( 'nocontentconvert' );
 650                 if( $mw->matchAndRemove( $text ) ) {
 651                         $this->mDoContentConvert = false;
 652                 }
 653
 654                 // no conversion if redirecting
 655                 $mw =& MagicWord::get( 'redirect' );
 656                 if( $mw->matchStart( $text ))
 657                         return $text;
 658
 659                 // for title convertion
 660                 if ($isTitle) return $this->convertTitle($text);
 661
 662                 $plang = $this->getPreferredVariant();
 663
 664                 $tarray = explode($this->mMarkup['begin'], $text);
 665                 $tfirst = array_shift($tarray);
 666                 if($this->mDoContentConvert)
 667                         $text = $this->autoConvert($tfirst,$plang);
 668                 else
 669                         $text = $tfirst;
 670                 foreach($tarray as $txt) {
 671                         $marked = explode($this->mMarkup['end'], $txt, 2);
 672
 673                         // strip the flags from syntax like -{T| ... }-
 674                         list($rules,$flags) = $this->parseFlags($marked[0]);
 675
 676                         $text .= $this->parseRules($rules,$flags,$plang);
 677
 678                         if(array_key_exists(1, $marked)){
 679                                 if( $this->mDoContentConvert )
 680                                         $text .= $this->autoConvert($marked[1],$plang);
 681                                 else
 682                                         $text .= $marked[1];
 683                         }
 684                 }
 685
 686                 return $text;
 687         }
 688
 689         /**
 690          * if a language supports multiple variants, it is
 691          * possible that non-existing link in one variant
 692          * actually exists in another variant. this function
 693          * tries to find it. See e.g. LanguageZh.php
 694          *
 695          * @param string $link the name of the link
 696          * @param mixed $nt the title object of the link
 697          * @return null the input parameters may be modified upon return
 698          * @access public
 699          */
 700         function findVariantLink( &$link, &$nt ) {
 701                 global $wgDisableLangConversion;
 702                 $linkBatch = new LinkBatch();
 703
 704                 $ns=NS_MAIN;
 705
 706                 if(is_object($nt))
 707                         $ns = $nt->getNamespace();
 708
 709                 $variants = $this->autoConvertToAllVariants($link);
 710                 if($variants == false) //give up
 711                         return;
 712
 713                 $titles = array();
 714
 715                 foreach( $variants as $v ) {
 716                         if($v != $link){
 717                                 $varnt = Title::newFromText( $v, $ns );
 718                                 if(!is_null($varnt)){
 719                                         $linkBatch->addObj($varnt);
 720                                         $titles[]=$varnt;
 721                                 }
 722                         }
 723                 }
 724
 725                 // fetch all variants in single query
 726                 $linkBatch->execute();
 727
 728                 foreach( $titles as $varnt ) {
 729                         if( $varnt->getArticleID() > 0 ) {
 730                                 $nt = $varnt;
 731                                 if( !$wgDisableLangConversion )
 732                                         $link = $v;
 733                                 break;
 734                         }
 735                 }
 736         }
 737
 738     /**
 739          * returns language specific hash options
 740          *
 741          * @access public
 742          */
 743         function getExtraHashOptions() {
 744                 $variant = $this->getPreferredVariant();
 745                 return '!' . $variant ;
 746         }
 747
 748     /**
 749          * get title text as defined in the body of the article text
 750          *
 751          * @access public
 752          */
 753         function getParsedTitle() {
 754                 return $this->mTitleDisplay;
 755         }
 756
 757         /**
 758          * a write lock to the cache
 759          *
 760          * @private
 761          */
 762         function lockCache() {
 763                 global $wgMemc;
 764                 $success = false;
 765                 for($i=0; $i<30; $i++) {
 766                         if($success = $wgMemc->add($this->mCacheKey . "lock", 1, 10))
 767                                 break;
 768                         sleep(1);
 769                 }
 770                 return $success;
 771         }
 772
 773         /**
 774          * unlock cache
 775          *
 776          * @private
 777          */
 778         function unlockCache() {
 779                 global $wgMemc;
 780                 $wgMemc->delete($this->mCacheKey . "lock");
 781         }
 782
 783
 784         /**
 785          * Load default conversion tables
 786          * This method must be implemented in derived class
 787          *
 788          * @private
 789          */
 790         function loadDefaultTables() {
 791                 $name = get_class($this);
 792                 wfDie("Must implement loadDefaultTables() method in class $name");
 793         }
 794
 795         /**
 796          * load conversion tables either from the cache or the disk
 797          * @private
 798          */
 799         function loadTables($fromcache=true) {
 800                 global $wgMemc;
 801                 if( $this->mTablesLoaded )
 802                         return;
 803                 wfProfileIn( __METHOD__ );
 804                 $this->mTablesLoaded = true;
 805                 $this->mTables = false;
 806                 if($fromcache) {
 807                         wfProfileIn( __METHOD__.'-cache' );
 808                         $this->mTables = $wgMemc->get( $this->mCacheKey );
 809                         wfProfileOut( __METHOD__.'-cache' );
 810                 }
 811                 if ( !$this->mTables || !isset( $this->mTables[self::CACHE_VERSION_KEY] ) ) {
 812                         wfProfileIn( __METHOD__.'-recache' );
 813                         // not in cache, or we need a fresh reload.
 814                         // we will first load the default tables
 815                         // then update them using things in MediaWiki:Zhconversiontable/*
 816                         $this->loadDefaultTables();
 817                         foreach($this->mVariants as $var) {
 818                                 $cached = $this->parseCachedTable($var);
 819                                 $this->mTables[$var]->mergeArray($cached);
 820                         }
 821
 822                         $this->postLoadTables();
 823                         $this->mTables[self::CACHE_VERSION_KEY] = true;
 824
 825                         if($this->lockCache()) {
 826                                 $wgMemc->set($this->mCacheKey, $this->mTables, 43200);
 827                                 $this->unlockCache();
 828                         }
 829                         wfProfileOut( __METHOD__.'-recache' );
 830                 }
 831                 wfProfileOut( __METHOD__ );
 832         }
 833
 834     /**
 835          * Hook for post processig after conversion tables are loaded
 836          *
 837          */
 838         function postLoadTables() {}
 839
 840     /**
 841          * Reload the conversion tables
 842          *
 843          * @private
 844          */
 845         function reloadTables() {
 846                 if($this->mTables)
 847                         unset($this->mTables);
 848                 $this->mTablesLoaded = false;
 849                 $this->loadTables(false);
 850         }
 851
 852
 853         /**
 854          * parse the conversion table stored in the cache
 855          *
 856          * the tables should be in blocks of the following form:
 857          *              -{
 858          *                      word => word ;
 859          *                      word => word ;
 860          *                      ...
 861          *              }-
 862          *
 863          *      to make the tables more manageable, subpages are allowed
 864          *      and will be parsed recursively if $recursive=true
 865          *
 866          */
 867         function parseCachedTable($code, $subpage='', $recursive=true) {
 868                 global $wgMessageCache;
 869                 static $parsed = array();
 870
 871                 if(!is_object($wgMessageCache))
 872                         return array();
 873
 874                 $key = 'Conversiontable/'.$code;
 875                 if($subpage)
 876                         $key .= '/' . $subpage;
 877
 878                 if(array_key_exists($key, $parsed))
 879                         return array();
 880
 881                 if ( strpos( $code, '/' ) === false ) {
 882                         $txt = $wgMessageCache->get( 'Conversiontable', true, $code );
 883                 } else {
 884                         $title = Title::makeTitleSafe( NS_MEDIAWIKI, "Conversiontable/$code" );
 885                         if ( $title && $title->exists() ) {
 886                                 $article = new Article( $title );
 887                                 $txt = $article->getContents();
 888                         } else {
 889                                 $txt = '';
 890                         }
 891                 }
 892
 893                 // get all subpage links of the form
 894                 // [[MediaWiki:conversiontable/zh-xx/...|...]]
 895                 $linkhead = $this->mLangObj->getNsText(NS_MEDIAWIKI) . ':Conversiontable';
 896                 $subs = explode('[[', $txt);
 897                 $sublinks = array();
 898                 foreach( $subs as $sub ) {
 899                         $link = explode(']]', $sub, 2);
 900                         if(count($link) != 2)
 901                                 continue;
 902                         $b = explode('|', $link[0]);
 903                         $b = explode('/', trim($b[0]), 3);
 904                         if(count($b)==3)
 905                                 $sublink = $b[2];
 906                         else
 907                                 $sublink = '';
 908
 909                         if($b[0] == $linkhead && $b[1] == $code) {
 910                                 $sublinks[] = $sublink;
 911                         }
 912                 }
 913
 914
 915                 // parse the mappings in this page
 916                 $blocks = explode($this->mMarkup['begin'], $txt);
 917                 array_shift($blocks);
 918                 $ret = array();
 919                 foreach($blocks as $block) {
 920                         $mappings = explode($this->mMarkup['end'], $block, 2);
 921                         $stripped = str_replace(array("'", '"', '*','#'), '', $mappings[0]);
 922                         $table = explode( ';', $stripped );
 923                         foreach( $table as $t ) {
 924                                 $m = explode( '=>', $t );
 925                                 if( count( $m ) != 2)
 926                                         continue;
 927                                 // trim any trailling comments starting with '//'
 928                                 $tt = explode('//', $m[1], 2);
 929                                 $ret[trim($m[0])] = trim($tt[0]);
 930                         }
 931                 }
 932                 $parsed[$key] = true;
 933
 934
 935                 // recursively parse the subpages
 936                 if($recursive) {
 937                         foreach($sublinks as $link) {
 938                                 $s = $this->parseCachedTable($code, $link, $recursive);
 939                                 $ret = array_merge($ret, $s);
 940                         }
 941                 }
 942
 943                 if ($this->mUcfirst) {
 944                         foreach ($ret as $k => $v) {
 945                                 $ret[Language::ucfirst($k)] = Language::ucfirst($v);
 946                         }
 947                 }
 948                 return $ret;
 949         }
 950
 951         /**
 952          * Enclose a string with the "no conversion" tag. This is used by
 953          * various functions in the Parser
 954          *
 955          * @param string $text text to be tagged for no conversion
 956          * @return string the tagged text
 957          * @public
 958          */
 959         function markNoConversion($text, $noParse=false) {
 960                 # don't mark if already marked
 961                 if(strpos($text, $this->mMarkup['begin']) ||
 962                    strpos($text, $this->mMarkup['end']))
 963                         return $text;
 964
 965                 $ret = $this->mMarkup['begin'] .'R|'. $text . $this->mMarkup['end'];
 966                 return $ret;
 967         }
 968
 969         /**
 970          * convert the sorting key for category links. this should make different
 971          * keys that are variants of each other map to the same key
 972          */
 973         function convertCategoryKey( $key ) {
 974                 return $key;
 975         }
 976         /**
 977          * hook to refresh the cache of conversion tables when
 978          * MediaWiki:conversiontable* is updated
 979          * @private
 980          */
 981         function OnArticleSaveComplete($article, $user, $text, $summary, $isminor, $iswatch, $section, $flags, $revision) {
 982                 $titleobj = $article->getTitle();
 983                 if($titleobj->getNamespace() == NS_MEDIAWIKI) {
 984                         $title = $titleobj->getDBkey();
 985                         $t = explode('/', $title, 3);
 986                         $c = count($t);
 987                         if( $c > 1 && $t[0] == 'Conversiontable' ) {
 988                                 if(in_array($t[1], $this->mVariants)) {
 989                                         $this->reloadTables();
 990                                 }
 991                         }
 992                 }
 993                 return true;
 994         }
 995
 996         /**
 997          * Armour rendered math against conversion
 998          * Wrap math into rawoutput -{R| math }- syntax
 999          * @public
1000          */
1001         function armourMath($text){
1002                 $ret = $this->mMarkup['begin'] . 'R|' . $text . $this->mMarkup['end'];
1003                 return $ret;
1004         }
1005 }