languages/LanguageConverter.php

   1 <?php
   2 /**
   3   * @package MediaWiki
   4   * @subpackage Language
   5   *
   6   * @author Zhengzhu Feng <zhengzhu@gmail.com>
   7   * @license http://www.gnu.org/copyleft/gpl.html GNU General Public License
   8   */
   9
  10 class LanguageConverter {
  11         var $mPreferredVariant='';
  12         var $mMainLanguageCode;
  13         var $mVariants, $mVariantFallbacks;
  14         var $mTablesLoaded = false;
  15         var $mTables;
  16         var $mTitleDisplay='';
  17         var $mDoTitleConvert=true, $mDoContentConvert=true;
  18         var $mCacheKey;
  19         var $mLangObj;
  20         var $mMarkup;
  21         /**
  22      * Constructor
  23          *
  24      * @param string $maincode the main language code of this language
  25      * @param array $variants the supported variants of this language
  26      * @param array $variantfallback the fallback language of each variant
  27      * @param array $markup array defining the markup used for manual conversion
  28      * @access public
  29      */
  30         function LanguageConverter($langobj, $maincode,
  31                                                                 $variants=array(),
  32                                                                 $variantfallbacks=array(),
  33                                                                 $markup=array('begin'=>'-{',
  34                                                                                           'codesep'=>':',
  35                                                                                           'varsep'=>';',
  36                                                                                           'end'=>'}-')) {
  37                 global $wgDBname;
  38                 $this->mLangObj = $langobj;
  39                 $this->mMainLanguageCode = $maincode;
  40                 $this->mVariants = $variants;
  41                 $this->mVariantFallbacks = $variantfallbacks;
  42                 $this->mCacheKey = $wgDBname . ":conversiontables";
  43                 $this->mMarkup = $markup;
  44         }
  45
  46         /**
  47      * @access public
  48      */
  49         function getVariants() {
  50                 return $this->mVariants;
  51         }
  52
  53         /**
  54          * in case some variant is not defined in the markup, we need
  55          * to have some fallback. for example, in zh, normally people
  56          * will define zh-cn and zh-tw, but less so for zh-sg or zh-hk.
  57          * when zh-sg is preferred but not defined, we will pick zh-cn
  58          * in this case. right now this is only used by zh.
  59          *
  60          * @param string $v the language code of the variant
  61          * @return string the code of the fallback language or false if there is no fallback
  62      * @access private
  63         */
  64         function getVariantFallback($v) {
  65                 return $this->mVariantFallbacks[$v];
  66         }
  67
  68
  69         /**
  70      * get preferred language variants.
  71      * @return string the preferred language code
  72      * @access public
  73         */
  74         function getPreferredVariant() {
  75                 global $wgUser, $wgRequest;
  76
  77                 if($this->mPreferredVariant)
  78                         return $this->mPreferredVariant;
  79
  80                 // see if the preference is set in the request
  81                 $req = $wgRequest->getText( 'variant' );
  82                 if( in_array( $req, $this->mVariants ) ) {
  83                         $this->mPreferredVariant = $req;
  84                         return $req;
  85                 }
  86
  87                 // get language variant preference from logged in users
  88                 if(is_object($wgUser) && $wgUser->isLoggedIn() )  {
  89                         $this->mPreferredVariant = $wgUser->getOption('variant');
  90                 }
  91
  92                 # FIXME rewrite code for parsing http header. The current code
  93                 # is written specific for detecting zh- variants
  94                 if( !$this->mPreferredVariant ) {
  95                         // see if some zh- variant is set in the http header,
  96                         $this->mPreferredVariant=$this->mMainLanguageCode;
  97                         if(array_key_exists('HTTP_ACCEPT_LANGUAGE', $_SERVER)) {
  98                                 $header = str_replace( '_', '-', strtolower($_SERVER["HTTP_ACCEPT_LANGUAGE"]));
  99                                 $zh = strstr($header, 'zh-');
 100                                 if($zh) {
 101                                         $this->mPreferredVariant = substr($zh,0,5);
 102                                 }
 103                         }
 104                 }
 105
 106                 return $this->mPreferredVariant;
 107         }
 108
 109         /**
 110      * dictionary-based conversion
 111      *
 112      * @param string $text the text to be converted
 113      * @param string $toVariant the target language code
 114      * @return string the converted text
 115      * @access private
 116      */
 117         function autoConvert($text, $toVariant=false) {
 118                 $fname="LanguageConverter::autoConvert";
 119                 wfProfileIn( $fname );
 120
 121                 if(!$this->mTablesLoaded)
 122                         $this->loadTables();
 123
 124                 if(!$toVariant)
 125                         $toVariant = $this->getPreferredVariant();
 126                 if(!in_array($toVariant, $this->mVariants))
 127                         return $text;
 128
 129                 $ret = '';
 130
 131                 $a = explode('<', $text);
 132                 $a0 = array_shift($a);
 133                 $ret .= strtr($a0, $this->mTables[$toVariant]);
 134                 foreach( $a as $aa ) {
 135                         $b = explode('>', $aa, 2);
 136                         $ret .= '<' . $b[0];
 137                         if(sizeof($b) == 2)
 138                                 $ret .= '>' . strtr($b[1], $this->mTables[$toVariant]);
 139                 }
 140
 141 #               /* put back the marker if any */
 142 #               if(!empty($reg)) {
 143 #                       $reg = '<'.$reg.'>';
 144 #                       $ret = preg_replace('/'.$reg.'/', '${1}', $ret);
 145 #               }
 146 #
 147                 wfProfileOut( $fname );
 148                 return $ret;
 149         }
 150
 151         /**
 152      * convert text to all supported variants
 153      *
 154      * @param string $text the text to be converted
 155      * @return array of string
 156      * @access private
 157      */
 158         function autoConvertToAllVariants($text) {
 159                 $fname="LanguageConverter::autoConvertToAllVariants";
 160                 wfProfileIn( $fname );
 161                 if( !$this->mTablesLoaded )
 162                         $this->loadTables();
 163
 164                 $ret = array();
 165                 foreach($this->mVariants as $variant) {
 166                         $ret[$variant] = strtr($text, $this->mTables[$variant]);
 167                 }
 168                 wfProfileOut( $fname );
 169                 return $ret;
 170         }
 171
 172         /**
 173          * convert text to different variants of a language. the automatic
 174          * conversion is done in autoConvert(). here we parse the text
 175          * marked with -{}-, which specifies special conversions of the
 176          * text that can not be accomplished in autoConvert()
 177          *
 178          * syntax of the markup:
 179          * -{code1:text1;code2:text2;...}-  or
 180          * -{text}- in which case no conversion should take place for text
 181      *
 182      * @param string $text text to be converted
 183      * @param bool $isTitle whether this conversion is for the article title
 184      * @return string converted text
 185      * @access public
 186      */
 187         function convert( $text , $isTitle=false) {
 188                 global $wgDisableLangConversion;
 189                 if($wgDisableLangConversion)
 190                         return $text;
 191
 192                 $mw =& MagicWord::get( MAG_NOTITLECONVERT );
 193                 if( $mw->matchAndRemove( $text ) )
 194                         $this->mDoTitleConvert = false;
 195
 196                 $mw =& MagicWord::get( MAG_NOCONTENTCONVERT );
 197                 if( $mw->matchAndRemove( $text ) ) {
 198                         $this->mDoContentConvert = false;
 199                 }
 200
 201                 // no conversion if redirecting
 202                 $mw =& MagicWord::get( MAG_REDIRECT );
 203                 if( $mw->matchStart( $text ))
 204                         return $text;
 205
 206                 if( $isTitle ) {
 207                         if( !$this->mDoTitleConvert ) {
 208                                 $this->mTitleDisplay = $text;
 209                                 return $text;
 210                         }
 211                         if( !empty($this->mTitleDisplay))
 212                                 return $this->mTitleDisplay;
 213
 214                         global $wgRequest;
 215                         $isredir = $wgRequest->getText( 'redirect', 'yes' );
 216                         $action = $wgRequest->getText( 'action' );
 217                         if ( $isredir == 'no' || $action == 'edit' ) {
 218                                 return $text;
 219                         }
 220                         else {
 221                                 $this->mTitleDisplay = $this->autoConvert($text);
 222                                 return $this->mTitleDisplay;
 223                         }
 224                 }
 225
 226                 if( !$this->mDoContentConvert )
 227                         return $text;
 228
 229                 $search = array('/('.UNIQ_PREFIX.'-[a-zA-Z0-9]+)/', //nowiki marker
 230                                         '/(&[a-z#][a-z0-9]+;)/', //html entities
 231                         );
 232                 $replace = $this->mMarkup['begin'].'${1}'.$this->mMarkup['end'];
 233
 234                 $text = preg_replace($search, $replace, $text);
 235
 236                 $plang = $this->getPreferredVariant();
 237                 $fallback = $this->mVariantFallbacks[$plang];
 238                 $tarray = explode($this->mMarkup['begin'], $text);
 239                 $tfirst = array_shift($tarray);
 240                 $text = $this->autoConvert($tfirst);
 241                 foreach($tarray as $txt) {
 242                         $marked = explode($this->mMarkup['end'], $txt);
 243
 244 # FIXME: this may cause trouble...
 245                         //strip &nbsp; since it interferes with the parsing, plus,
 246                         //all spaces should be stripped in this tag anyway.
 247                         $marked[0] = str_replace('&nbsp;', '', $marked[0]);
 248
 249                         /* see if this conversion has special meaning
 250                            # for article title:
 251                                  -{T|zh-cn:foo;zh-tw:bar}-
 252                            # convert all occurence of foo/bar in this article:
 253                                  -{A|zh-cn:foo;zh-tw:bar}-
 254                         */
 255                         $flag = '';
 256                         $choice = false;
 257                         $tt = explode("|", $marked[0], 2);
 258                         if(sizeof($tt) == 2) {
 259                                 $flag = trim($tt[0]);
 260                                 $choice = explode(";", $tt[1]);
 261                         }
 262
 263                         if(!$choice) {
 264                                 $choice = explode($this->mMarkup['varsep'], $marked[0]);
 265                         }
 266                         $disp = '';
 267                         $carray = array();
 268                         if(!array_key_exists(1, $choice)) {
 269                                 /* a single choice */
 270                                 $disp = $choice[0];
 271
 272                                 /* fill the carray if the conversion is for the whole article*/
 273                                 if($flag == 'A') {
 274                                         foreach($this->mVariants as $v) {
 275                                                 $carray[$v] = $disp;
 276                                         }
 277                                 }
 278                         }
 279                         else {
 280                                 foreach($choice as $c) {
 281                                         $v = explode($this->mMarkup['codesep'], $c);
 282                                         if(sizeof($v) != 2) // syntax error, skip
 283                                                 continue;
 284                                         $carray[trim($v[0])] = trim($v[1]);
 285                                 }
 286                                 if(array_key_exists($plang, $carray))
 287                                         $disp = $carray[$plang];
 288                                 else if(array_key_exists($fallback, $carray))
 289                                         $disp = $carray[$fallback];
 290                         }
 291                         if(empty($disp)) { // syntax error
 292                                 $text .= $marked[0];
 293                         }
 294                         else {
 295                                 if($flag == 'T') // for title only
 296                                         $this->mTitleDisplay = $disp;
 297                                 else {
 298                                         $text .= $disp;
 299                                         if($flag == 'A') {
 300                                                 /* modify the conversion table for this session*/
 301
 302                                                 /* fill in the missing variants, if any,
 303                                                     with fallbacks */
 304                                                 foreach($this->mVariants as $v) {
 305                                                         if(!array_key_exists($v, $carray)) {
 306                                                                 $vf = $this->getVariantFallback($v);
 307                                                                 if(array_key_exists($vf, $carray))
 308                                                                         $carray[$v] = $carray[$vf];
 309                                                         }
 310                                                 }
 311                                                 foreach($this->mVariants as $vfrom) {
 312                                                         if(!array_key_exists($vfrom, $carray))
 313                                                                 continue;
 314                                                         foreach($this->mVariants as $vto) {
 315                                                                 if($vfrom == $vto)
 316                                                                         continue;
 317                                                                 if(!array_key_exists($vto, $carray))
 318                                                                         continue;
 319                                                                 $this->mTables[$vto][$carray[$vfrom]] = $carray[$vto];
 320                                                         }
 321                                                 }
 322                                         }
 323                                 }
 324                         }
 325                         if(array_key_exists(1, $marked))
 326                                 $text .= $this->autoConvert($marked[1]);
 327                 }
 328
 329                 return $text;
 330         }
 331
 332
 333         /**
 334          * if a language supports multiple variants, it is
 335          * possible that non-existing link in one variant
 336          * actually exists in another variant. this function
 337          * tries to find it. See e.g. LanguageZh.php
 338          *
 339          * @param string $link the name of the link
 340          * @param mixed $nt the title object of the link
 341          * @return null the input parameters may be modified upon return
 342      * @access public
 343          */
 344         function findVariantLink( &$link, &$nt ) {
 345                 static $count=0; //used to limit this operation
 346                 static $cache=array();
 347                 global $wgDisableLangConversion;
 348                 $pref = $this->getPreferredVariant();
 349                 if( $count > 50 )
 350                         return;
 351                 $count++;
 352                 $variants = $this->autoConvertToAllVariants($link);
 353                 if($variants == false) //give up
 354                         return;
 355                 foreach( $variants as $v ) {
 356                         if(isset($cache[$v]))
 357                                 continue;
 358                         $cache[$v] = 1;
 359                         $varnt = Title::newFromText( $v );
 360                         if( $varnt && $varnt->getArticleID() > 0 ) {
 361                                 $nt = $varnt;
 362                                 if( !$wgDisableLangConversion && $pref != 'zh' )
 363                                         $link = $v;
 364                                 break;
 365                         }
 366                 }
 367         }
 368
 369     /**
 370      * returns language specific hash options
 371      *
 372      * @access public
 373      */
 374         function getExtraHashOptions() {
 375                 $variant = $this->getPreferredVariant();
 376                 return '!' . $variant ;
 377         }
 378
 379     /**
 380      * get title text as defined in the body of the article text
 381      *
 382      * @access public
 383      */
 384         function getParsedTitle() {
 385                 return $this->mTitleDisplay;
 386         }
 387
 388         /**
 389      * a write lock to the cache
 390      *
 391      * @access private
 392      */
 393         function lockCache() {
 394                 global $wgMemc;
 395                 $success = false;
 396                 for($i=0; $i<30; $i++) {
 397                         if($success = $wgMemc->add($this->mCacheKey . "lock", 1, 10))
 398                                 break;
 399                         sleep(1);
 400                 }
 401                 return $success;
 402         }
 403
 404         /**
 405      * unlock cache
 406      *
 407      * @access private
 408      */
 409         function unlockCache() {
 410                 global $wgMemc;
 411                 $wgMemc->delete($this->mCacheKey . "lock");
 412         }
 413
 414
 415         /**
 416      * Load default conversion tables
 417      * This method must be implemented in derived class
 418      *
 419      * @access private
 420      */
 421         function loadDefaultTables() {
 422                 $name = get_class($this);
 423                 die("Must implement loadDefaultTables() method in class $name");
 424         }
 425
 426         /**
 427      * load conversion tables either from the cache or the disk
 428      * @access private
 429      */
 430         function loadTables($fromcache=true) {
 431                 global $wgMemc;
 432                 if( $this->mTablesLoaded )
 433                         return;
 434                 $this->mTablesLoaded = true;
 435                 if($fromcache) {
 436                         $this->mTables = $wgMemc->get( $this->mCacheKey );
 437                         if( !empty( $this->mTables ) ) //all done
 438                                 return;
 439                 }
 440                 // not in cache, or we need a fresh reload.
 441                 // we will first load the default tables
 442                 // then update them using things in MediaWiki:Zhconversiontable/*
 443                 global $wgMessageCache;
 444                 $this->loadDefaultTables();
 445                 foreach($this->mVariants as $var) {
 446                         $cached = $this->parseCachedTable($var);
 447                         $this->mTables[$var] = array_merge($this->mTables[$var], $cached);
 448                 }
 449
 450                 $this->postLoadTables();
 451
 452                 if($this->lockCache()) {
 453                         $wgMemc->set($this->mCacheKey, $this->mTables, 43200);
 454                         $this->unlockCache();
 455                 }
 456         }
 457
 458     /**
 459      * Hook for post processig after conversion tables are loaded
 460      *
 461      */
 462         function postLoadTables() {}
 463
 464         /* deprecated? */
 465         function updateTablexxxx($code, $table) {
 466                 global $wgMemc;
 467                 if(!$this->mTablesLoaded)
 468                         $this->loadTables();
 469
 470                 $this->mTables[$code] = array_merge($this->mTables[$code], $table);
 471                 if($this->lockCache()) {
 472                         $wgMemc->delete($this->mCacheKey);
 473                         $wgMemc->set($this->mCacheKey, $this->mTables, 43200);
 474                         $this->unlockCache();
 475                 }
 476         }
 477
 478     /**
 479      * Reload the conversion tables
 480      *
 481      * @access private
 482      */
 483         function reloadTables() {
 484                 if($this->mTables)
 485                         unset($this->mTables);
 486                 $this->mTablesLoaded = false;
 487                 $this->loadTables(false);
 488         }
 489
 490
 491         /**
 492      * parse the conversion table stored in the cache
 493      *
 494      * the tables should be in blocks of the following form:
 495
 496      *          -{
 497      *                  word => word ;
 498      *                  word => word ;
 499      *                  ...
 500      *          }-
 501      *
 502      *  to make the tables more manageable, subpages are allowed
 503      *  and will be parsed recursively if $recursive=true
 504      *
 505      * @access private
 506          */
 507         function parseCachedTable($code, $subpage='', $recursive=true) {
 508                 global $wgMessageCache;
 509                 static $parsed = array();
 510
 511                 if(!is_object($wgMessageCache))
 512                         return array();
 513
 514                 $key = 'Conversiontable/'.$code;
 515                 if($subpage)
 516                         $key .= '/' . $subpage;
 517
 518                 if(array_key_exists($key, $parsed))
 519                         return array();
 520
 521
 522                 $txt = $wgMessageCache->get( $key, true, true, true );
 523
 524                 // get all subpage links of the form
 525                 // [[MediaWiki:conversiontable/zh-xx/...|...]]
 526                 $linkhead = $this->mLangObj->getNsText(NS_MEDIAWIKI) . ':Conversiontable';
 527                 $subs = explode('[[', $txt);
 528                 $sublinks = array();
 529                 foreach( $subs as $sub ) {
 530                         $link = explode(']]', $sub, 2);
 531                         if(count($link) != 2)
 532                                 continue;
 533                         $b = explode('|', $link[0]);
 534                         $b = explode('/', trim($b[0]), 3);
 535                         if(count($b)==3)
 536                                 $sublink = $b[2];
 537                         else
 538                                 $sublink = '';
 539
 540                         if($b[0] == $linkhead && $b[1] == $code) {
 541                                 $sublinks[] = $sublink;
 542                         }
 543                 }
 544
 545
 546                 // parse the mappings in this page
 547                 $blocks = explode('-{', $txt);
 548                 array_shift($blocks);
 549                 $ret = array();
 550                 foreach($blocks as $block) {
 551                         $mappings = explode('}-', $block, 2);
 552                         $stripped = str_replace(array("'", '"', '*','#'), '', $mappings[0]);
 553                         $table = explode( ';', $stripped );
 554                         foreach( $table as $t ) {
 555                                 $m = explode( '=>', $t );
 556                                 if( count( $m ) != 2)
 557                                         continue;
 558                                 // trim any trailling comments starting with '//'
 559                                 $tt = explode('//', $m[1], 2);
 560                                 $ret[trim($m[0])] = trim($tt[0]);
 561                         }
 562                 }
 563                 $parsed[$key] = true;
 564
 565
 566                 // recursively parse the subpages
 567                 if($recursive) {
 568                         foreach($sublinks as $link) {
 569                                 $s = $this->parseCachedTable($code, $link, $recursive);
 570                                 $ret = array_merge($ret, $s);
 571                         }
 572                 }
 573                 return $ret;
 574         }
 575
 576         /**
 577          * Enclose a string with the "no conversion" tag. This is used by
 578          * various functions in the Parser
 579          *
 580          * @param string $text text to be tagged for no conversion
 581          * @return string the tagged text
 582         */
 583         function markNoConversion($text) {
 584                 $ret = $this->mMarkup['begin'] . $text . $this->mMarkup['end'];
 585                 return $ret;
 586         }
 587
 588         /**
 589      * hook to refresh the cache of conversion tables when
 590      * MediaWiki:conversiontable* is updated
 591      * @access private
 592         */
 593         function OnArticleSaveComplete($article, $user, $text, $summary, $isminor, $iswatch, $section) {
 594                 $titleobj = $article->getTitle();
 595                 if($titleobj->getNamespace() == NS_MEDIAWIKI) {
 596             /*
 597                         global $wgContLang; // should be an LanguageZh.
 598                         if(get_class($wgContLang) != 'languagezh')
 599                                 return true;
 600             */
 601                         $title = $titleobj->getDBkey();
 602                         $t = explode('/', $title, 3);
 603                         $c = count($t);
 604                         if( $c > 1 && $t[0] == 'Conversiontable' ) {
 605                                 if(in_array($t[1], $this->mVariants)) {
 606                                         $this->reloadTables();
 607                                 }
 608                         }
 609                 }
 610                 return true;
 611         }
 612 }
 613
 614 ?>