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                         //strip &nbsp; since it interferes with the parsing, plus,
 245                         //all spaces should be stripped in this tag anyway.
 246                         $marked[0] = str_replace('&nbsp;', '', $marked[0]);
 247
 248                         /* see if this conversion has special meaning
 249                            # for article title:
 250                                  -{T|zh-cn:foo;zh-tw:bar}-
 251                            # convert all occurence of foo/bar in this article:
 252                                  -{A|zh-cn:foo;zh-tw:bar}-
 253                         */
 254                         $flag = '';
 255                         $choice = false;
 256                         $tt = explode("|", $marked[0], 2);
 257                         if(sizeof($tt) == 2) {
 258                                 $flag = trim($tt[0]);
 259                                 $choice = explode(";", $tt[1]);
 260                         }
 261
 262                         if(!$choice) {
 263                                 $choice = explode($this->mMarkup['varsep'], $marked[0]);
 264                         }
 265                         $disp = '';
 266                         $carray = array();
 267                         if(!array_key_exists(1, $choice)) {
 268                                 /* a single choice */
 269                                 $disp = $choice[0];
 270
 271                                 /* fill the carray if the conversion is for the whole article*/
 272                                 if($flag == 'A') {
 273                                         foreach($this->mVariants as $v) {
 274                                                 $carray[$v] = $disp;
 275                                         }
 276                                 }
 277                         }
 278                         else {
 279                                 foreach($choice as $c) {
 280                                         $v = explode($this->mMarkup['codesep'], $c);
 281                                         if(sizeof($v) != 2) // syntax error, skip
 282                                                 continue;
 283                                         $carray[trim($v[0])] = trim($v[1]);
 284                                 }
 285                                 if(array_key_exists($plang, $carray))
 286                                         $disp = $carray[$plang];
 287                                 else if(array_key_exists($fallback, $carray))
 288                                         $disp = $carray[$fallback];
 289                         }
 290                         if(empty($disp)) { // syntax error
 291                                 $text .= $marked[0];
 292                         }
 293                         else {
 294                                 if($flag == 'T') // for title only
 295                                         $this->mTitleDisplay = $disp;
 296                                 else {
 297                                         $text .= $disp;
 298                                         if($flag == 'A') {
 299                                                 /* modify the conversion table for this session*/
 300
 301                                                 /* fill in the missing variants, if any,
 302                                                     with fallbacks */
 303                                                 foreach($this->mVariants as $v) {
 304                                                         if(!array_key_exists($v, $carray)) {
 305                                                                 $vf = $this->getVariantFallback($v);
 306                                                                 if(array_key_exists($vf, $carray))
 307                                                                         $carray[$v] = $carray[$vf];
 308                                                         }
 309                                                 }
 310                                                 foreach($this->mVariants as $vfrom) {
 311                                                         if(!array_key_exists($vfrom, $carray))
 312                                                                 continue;
 313                                                         foreach($this->mVariants as $vto) {
 314                                                                 if($vfrom == $vto)
 315                                                                         continue;
 316                                                                 if(!array_key_exists($vto, $carray))
 317                                                                         continue;
 318                                                                 $this->mTables[$vto][$carray[$vfrom]] = $carray[$vto];
 319                                                         }
 320                                                 }
 321                                         }
 322                                 }
 323                         }
 324                         if(array_key_exists(1, $marked))
 325                                 $text .= $this->autoConvert($marked[1]);
 326                 }
 327
 328                 return $text;
 329         }
 330
 331
 332         /**
 333          * if a language supports multiple variants, it is
 334          * possible that non-existing link in one variant
 335          * actually exists in another variant. this function
 336          * tries to find it. See e.g. LanguageZh.php
 337          *
 338          * @param string $link the name of the link
 339          * @param mixed $nt the title object of the link
 340          * @return null the input parameters may be modified upon return
 341      * @access public
 342          */
 343         function findVariantLink( &$link, &$nt ) {
 344                 static $count=0; //used to limit this operation
 345                 static $cache=array();
 346                 global $wgDisableLangConversion;
 347                 $pref = $this->getPreferredVariant();
 348                 if( $count > 50 )
 349                         return;
 350                 $count++;
 351                 $variants = $this->autoConvertToAllVariants($link);
 352                 if($variants == false) //give up
 353                         return;
 354                 foreach( $variants as $v ) {
 355                         if(isset($cache[$v]))
 356                                 continue;
 357                         $cache[$v] = 1;
 358                         $varnt = Title::newFromText( $v );
 359                         if( $varnt && $varnt->getArticleID() > 0 ) {
 360                                 $nt = $varnt;
 361                                 if( !$wgDisableLangConversion && $pref != 'zh' )
 362                                         $link = $v;
 363                                 break;
 364                         }
 365                 }
 366         }
 367
 368     /**
 369      * returns language specific hash options
 370      *
 371      * @access public
 372      */
 373         function getExtraHashOptions() {
 374                 $variant = $this->getPreferredVariant();
 375                 return '!' . $variant ;
 376         }
 377
 378     /**
 379      * get title text as defined in the body of the article text
 380      *
 381      * @access public
 382      */
 383         function getParsedTitle() {
 384                 return $this->mTitleDisplay;
 385         }
 386
 387         /**
 388      * a write lock to the cache
 389      *
 390      * @access private
 391      */
 392         function lockCache() {
 393                 global $wgMemc;
 394                 $success = false;
 395                 for($i=0; $i<30; $i++) {
 396                         if($success = $wgMemc->add($this->mCacheKey . "lock", 1, 10))
 397                                 break;
 398                         sleep(1);
 399                 }
 400                 return $success;
 401         }
 402
 403         /**
 404      * unlock cache
 405      *
 406      * @access private
 407      */
 408         function unlockCache() {
 409                 global $wgMemc;
 410                 $wgMemc->delete($this->mCacheKey . "lock");
 411         }
 412
 413
 414         /**
 415      * Load default conversion tables
 416      * This method must be implemented in derived class
 417      *
 418      * @access private
 419      */
 420         function loadDefaultTables() {
 421                 $name = get_class($this);
 422                 die("Must implement loadDefaultTables() method in class $name");
 423         }
 424
 425         /**
 426      * load conversion tables either from the cache or the disk
 427      * @access private
 428      */
 429         function loadTables($fromcache=true) {
 430                 global $wgMemc;
 431                 if( $this->mTablesLoaded )
 432                         return;
 433                 $this->mTablesLoaded = true;
 434                 if($fromcache) {
 435                         $this->mTables = $wgMemc->get( $this->mCacheKey );
 436                         if( !empty( $this->mTables ) ) //all done
 437                                 return;
 438                 }
 439                 // not in cache, or we need a fresh reload.
 440                 // we will first load the default tables
 441                 // then update them using things in MediaWiki:Zhconversiontable/*
 442                 global $wgMessageCache;
 443                 $this->loadDefaultTables();
 444                 foreach($this->mVariants as $var) {
 445                         $cached = $this->parseCachedTable($var);
 446                         $this->mTables[$var] = array_merge($this->mTables[$var], $cached);
 447                 }
 448
 449                 $this->postLoadTables();
 450
 451                 if($this->lockCache()) {
 452                         $wgMemc->set($this->mCacheKey, $this->mTables, 43200);
 453                         $this->unlockCache();
 454                 }
 455         }
 456
 457     /**
 458      * Hook for post processig after conversion tables are loaded
 459      *
 460      */
 461         function postLoadTables() {}
 462
 463         /* deprecated? */
 464         function updateTablexxxx($code, $table) {
 465                 global $wgMemc;
 466                 if(!$this->mTablesLoaded)
 467                         $this->loadTables();
 468
 469                 $this->mTables[$code] = array_merge($this->mTables[$code], $table);
 470                 if($this->lockCache()) {
 471                         $wgMemc->delete($this->mCacheKey);
 472                         $wgMemc->set($this->mCacheKey, $this->mTables, 43200);
 473                         $this->unlockCache();
 474                 }
 475         }
 476
 477     /**
 478      * Reload the conversion tables
 479      *
 480      * @access private
 481      */
 482         function reloadTables() {
 483                 if($this->mTables)
 484                         unset($this->mTables);
 485                 $this->mTablesLoaded = false;
 486                 $this->loadTables(false);
 487         }
 488
 489
 490         /**
 491      * parse the conversion table stored in the cache
 492      *
 493      * the tables should be in blocks of the following form:
 494
 495      *          -{
 496      *                  word => word ;
 497      *                  word => word ;
 498      *                  ...
 499      *          }-
 500      *
 501      *  to make the tables more manageable, subpages are allowed
 502      *  and will be parsed recursively if $recursive=true
 503      *
 504      * @access private
 505          */
 506         function parseCachedTable($code, $subpage='', $recursive=true) {
 507                 global $wgMessageCache;
 508                 static $parsed = array();
 509
 510                 if(!is_object($wgMessageCache))
 511                         return array();
 512
 513                 $key = 'Conversiontable/'.$code;
 514                 if($subpage)
 515                         $key .= '/' . $subpage;
 516
 517                 if(array_key_exists($key, $parsed))
 518                         return array();
 519
 520
 521                 $txt = $wgMessageCache->get( $key, true, true, true );
 522
 523                 // get all subpage links of the form
 524                 // [[MediaWiki:conversiontable/zh-xx/...|...]]
 525                 $linkhead = $this->mLangObj->getNsText(NS_MEDIAWIKI) . ':Conversiontable';
 526                 $subs = explode('[[', $txt);
 527                 $sublinks = array();
 528                 foreach( $subs as $sub ) {
 529                         $link = explode(']]', $sub, 2);
 530                         if(count($link) != 2)
 531                                 continue;
 532                         $b = explode('|', $link[0]);
 533                         $b = explode('/', trim($b[0]), 3);
 534                         if(count($b)==3)
 535                                 $sublink = $b[2];
 536                         else
 537                                 $sublink = '';
 538
 539                         if($b[0] == $linkhead && $b[1] == $code) {
 540                                 $sublinks[] = $sublink;
 541                         }
 542                 }
 543
 544
 545                 // parse the mappings in this page
 546                 $blocks = explode('-{', $txt);
 547                 array_shift($blocks);
 548                 $ret = array();
 549                 foreach($blocks as $block) {
 550                         $mappings = explode('}-', $block, 2);
 551                         $stripped = str_replace(array("'", '"', '*','#'), '', $mappings[0]);
 552                         $table = explode( ';', $stripped );
 553                         foreach( $table as $t ) {
 554                                 $m = explode( '=>', $t );
 555                                 if( count( $m ) != 2)
 556                                         continue;
 557                                 // trim any trailling comments starting with '//'
 558                                 $tt = explode('//', $m[1], 2);
 559                                 $ret[trim($m[0])] = trim($tt[0]);
 560                         }
 561                 }
 562                 $parsed[$key] = true;
 563
 564
 565                 // recursively parse the subpages
 566                 if($recursive) {
 567                         foreach($sublinks as $link) {
 568                                 $s = $this->parseCachedTable($code, $link, $recursive);
 569                                 $ret = array_merge($ret, $s);
 570                         }
 571                 }
 572                 return $ret;
 573         }
 574
 575         /**
 576          * Enclose a string with the "no conversion" tag. This is used by
 577          * various functions in the Parser
 578          *
 579          * @param string $text text to be tagged for no conversion
 580          * @return string the tagged text
 581         */
 582         function markNoConversion($text) {
 583                 $ret = $this->mMarkup['begin'] . $text . $this->mMarkup['end'];
 584         }
 585
 586         /**
 587      * hook to refresh the cache of conversion tables when
 588      * MediaWiki:conversiontable* is updated
 589      * @access private
 590         */
 591         function OnArticleSaveComplete($article, $user, $text, $summary, $isminor, $iswatch, $section) {
 592                 $titleobj = $article->getTitle();
 593                 if($titleobj->getNamespace() == NS_MEDIAWIKI) {
 594             /*
 595                         global $wgContLang; // should be an LanguageZh.
 596                         if(get_class($wgContLang) != 'languagezh')
 597                                 return true;
 598             */
 599                         $title = $titleobj->getDBkey();
 600                         $t = explode('/', $title, 3);
 601                         $c = count($t);
 602                         if( $c > 1 && $t[0] == 'Conversiontable' ) {
 603                                 if(in_array($t[1], $this->mVariants)) {
 604                                         $this->reloadTables();
 605                                 }
 606                         }
 607                 }
 608                 return true;
 609         }
 610 }
 611
 612 ?>