languages/LanguageConverter.php

   1 <?php
   2 /*
   3         Copyright (C) 2005 by Zhengzhu Feng (zhengzhu@gmail.com)
   4
   5         This file is distributed under the terms of GPL
   6
   7 */
   8
   9 /**
  10   * @package MediaWiki
  11   * @subpackage Language
  12   */
  13
  14 class LanguageConverter {
  15         var $mPreferredVariant='';
  16         var $mMainLanguageCode;
  17         var $mVariants, $mVariantFallbacks;
  18         var $mTablesLoaded = false;
  19         var $mTables;
  20         var $mTitleDisplay='';
  21         var $mDoTitleConvert=true, $mDoContentConvert=true;
  22         var $mCacheKey;
  23         var $mLangObj;
  24         var $mMarkup;
  25         /**
  26      * Constructor
  27          *
  28      * @param string $maincode the main language code of this language
  29      * @param array $variants the supported variants of this language
  30      * @param array $variantfallback the fallback language of each variant
  31      * @param array $markup array defining the markup used for manual conversion
  32      * @access public
  33      */
  34         function LanguageConverter($langobj, $maincode,
  35                                                                 $variants=array(),
  36                                                                 $variantfallbacks=array(),
  37                                                                 $markup=array('begin'=>'-{',
  38                                                                                           'codesep'=>':',
  39                                                                                           'varsep'=>';',
  40                                                                                           'end'=>'}-')) {
  41                 global $wgDBname;
  42                 $this->mLangObj = $langobj;
  43                 $this->mMainLanguageCode = $maincode;
  44                 $this->mVariants = $variants;
  45                 $this->mVariantFallbacks = $variantfallbacks;
  46                 $this->mCacheKey = $wgDBname . ":conversiontables";
  47                 $this->mMarkup = $markup;
  48         }
  49
  50         /**
  51      * @access public
  52      */
  53         function getVariants() {
  54                 return $this->mVariants;
  55         }
  56
  57         /**
  58          * in case some variant is not defined in the markup, we need
  59          * to have some fallback. for example, in zh, normally people
  60          * will define zh-cn and zh-tw, but less so for zh-sg or zh-hk.
  61          * when zh-sg is preferred but not defined, we will pick zh-cn
  62          * in this case. right now this is only used by zh.
  63          *
  64          * @param string $v the language code of the variant
  65          * @return string the code of the fallback language or false if there is no fallback
  66      * @access private
  67         */
  68         function getVariantFallback($v) {
  69                 return $this->mVariantFallbacks[$v];
  70         }
  71
  72
  73         /**
  74      * get preferred language variants.
  75      * @return string the preferred language code
  76      * @access public
  77         */
  78         function getPreferredVariant() {
  79                 global $wgUser, $wgRequest;
  80
  81                 if($this->mPreferredVariant)
  82                         return $this->mPreferredVariant;
  83
  84                 // see if the preference is set in the request
  85                 $req = $wgRequest->getText( 'variant' );
  86                 if( in_array( $req, $this->mVariants ) ) {
  87                         $this->mPreferredVariant = $req;
  88                         return $req;
  89                 }
  90
  91                 // get language variant preference from logged in users
  92                 if(is_object($wgUser) && $wgUser->isLoggedIn() )  {
  93                         $this->mPreferredVariant = $wgUser->getOption('variant');
  94                 }
  95
  96                 # FIXME rewrite code for parsing http header. The current code
  97                 # is written specific for detecting zh- variants
  98                 if( !$this->mPreferredVariant ) {
  99                         // see if some zh- variant is set in the http header,
 100                         $this->mPreferredVariant=$this->mMainLanguageCode;
 101                         if(array_key_exists('HTTP_ACCEPT_LANGUAGE', $_SERVER)) {
 102                                 $header = str_replace( '_', '-', strtolower($_SERVER["HTTP_ACCEPT_LANGUAGE"]));
 103                                 $zh = strstr($header, 'zh-');
 104                                 if($zh) {
 105                                         $this->mPreferredVariant = substr($zh,0,5);
 106                                 }
 107                         }
 108                 }
 109
 110                 return $this->mPreferredVariant;
 111         }
 112
 113         /**
 114      * dictionary-based conversion
 115      *
 116      * @param string $text the text to be converted
 117      * @param string $toVariant the target language code
 118      * @return string the converted text
 119      * @access private
 120      */
 121         function autoConvert($text, $toVariant=false) {
 122                 $fname="LanguageConverter::autoConvert";
 123                 wfProfileIn( $fname );
 124
 125                 if(!$this->mTablesLoaded)
 126                         $this->loadTables();
 127
 128                 if(!$toVariant)
 129                         $toVariant = $this->getPreferredVariant();
 130                 if(!in_array($toVariant, $this->mVariants))
 131                         return $text;
 132
 133                 $ret = '';
 134
 135                 $a = explode('<', $text);
 136                 $a0 = array_shift($a);
 137                 $ret .= strtr($a0, $this->mTables[$toVariant]);
 138                 foreach( $a as $aa ) {
 139                         $b = explode('>', $aa, 2);
 140                         $ret .= '<' . $b[0];
 141                         if(sizeof($b) == 2)
 142                                 $ret .= '>' . strtr($b[1], $this->mTables[$toVariant]);
 143                 }
 144
 145 #               /* put back the marker if any */
 146 #               if(!empty($reg)) {
 147 #                       $reg = '<'.$reg.'>';
 148 #                       $ret = preg_replace('/'.$reg.'/', '${1}', $ret);
 149 #               }
 150 #
 151                 wfProfileOut( $fname );
 152                 return $ret;
 153         }
 154
 155         /**
 156      * convert text to all supported variants
 157      *
 158      * @param string $text the text to be converted
 159      * @return array of string
 160      * @access private
 161      */
 162         function autoConvertToAllVariants($text) {
 163                 $fname="LanguageConverter::autoConvertToAllVariants";
 164                 wfProfileIn( $fname );
 165                 if( !$this->mTablesLoaded )
 166                         $this->loadTables();
 167
 168                 $ret = array();
 169                 foreach($this->mVariants as $variant) {
 170                         $ret[$variant] = strtr($text, $this->mTables[$variant]);
 171                 }
 172                 wfProfileOut( $fname );
 173                 return $ret;
 174         }
 175
 176         /**
 177          * convert text to different variants of a language. the automatic
 178          * conversion is done in autoConvert(). here we parse the text
 179          * marked with -{}-, which specifies special conversions of the
 180          * text that can not be accomplished in autoConvert()
 181          *
 182          * syntax of the markup:
 183          * -{code1:text1;code2:text2;...}-  or
 184          * -{text}- in which case no conversion should take place for text
 185      *
 186      * @param string $text text to be converted
 187      * @param bool $isTitle whether this conversion is for the article title
 188      * @return string converted text
 189      * @access public
 190      */
 191         function convert( $text , $isTitle=false) {
 192                 global $wgDisableLangConversion;
 193                 if($wgDisableLangConversion)
 194                         return $text;
 195
 196                 $mw =& MagicWord::get( MAG_NOTITLECONVERT );
 197                 if( $mw->matchAndRemove( $text ) )
 198                         $this->mDoTitleConvert = false;
 199
 200                 $mw =& MagicWord::get( MAG_NOCONTENTCONVERT );
 201                 if( $mw->matchAndRemove( $text ) ) {
 202                         $this->mDoContentConvert = false;
 203                 }
 204
 205                 // no conversion if redirecting
 206                 $mw =& MagicWord::get( MAG_REDIRECT );
 207                 if( $mw->matchStart( $text ))
 208                         return $text;
 209
 210                 if( $isTitle ) {
 211                         if( !$this->mDoTitleConvert ) {
 212                                 $this->mTitleDisplay = $text;
 213                                 return $text;
 214                         }
 215                         if( !empty($this->mTitleDisplay))
 216                                 return $this->mTitleDisplay;
 217
 218                         global $wgRequest;
 219                         $isredir = $wgRequest->getText( 'redirect', 'yes' );
 220                         $action = $wgRequest->getText( 'action' );
 221                         if ( $isredir == 'no' || $action == 'edit' ) {
 222                                 return $text;
 223                         }
 224                         else {
 225                                 $this->mTitleDisplay = $this->autoConvert($text);
 226                                 return $this->mTitleDisplay;
 227                         }
 228                 }
 229
 230                 if( !$this->mDoContentConvert )
 231                         return $text;
 232
 233                 $search = array('/('.UNIQ_PREFIX.'-[a-zA-Z0-9]+)/', //nowiki marker
 234                                         '/(&[a-z#][a-z0-9]+;)/', //html entities
 235                         );
 236                 $replace = $this->mMarkup['begin'].'${1}'.$this->mMarkup['end'];
 237
 238                 $text = preg_replace($search, $replace, $text);
 239
 240                 $plang = $this->getPreferredVariant();
 241                 $fallback = $this->mVariantFallbacks[$plang];
 242                 $tarray = explode($this->mMarkup['begin'], $text);
 243                 $tfirst = array_shift($tarray);
 244                 $text = $this->autoConvert($tfirst);
 245                 foreach($tarray as $txt) {
 246                         $marked = explode($this->mMarkup['end'], $txt);
 247
 248                         //strip &nbsp; since it interferes with the parsing, plus,
 249                         //all spaces should be stripped in this tag anyway.
 250                         $marked[0] = str_replace('&nbsp;', '', $marked[0]);
 251
 252                         /* see if this conversion has special meaning
 253                            # for article title:
 254                                  -{T|zh-cn:foo;zh-tw:bar}-
 255                            # convert all occurence of foo/bar in this article:
 256                                  -{A|zh-cn:foo;zh-tw:bar}-
 257                         */
 258                         $flag = '';
 259                         $choice = false;
 260                         $tt = explode("|", $marked[0], 2);
 261                         if(sizeof($tt) == 2) {
 262                                 $flag = trim($tt[0]);
 263                                 $choice = explode(";", $tt[1]);
 264                         }
 265
 266                         if(!$choice) {
 267                                 $choice = explode($this->mMarkup['varsep'], $marked[0]);
 268                         }
 269                         $disp = '';
 270                         $carray = array();
 271                         if(!array_key_exists(1, $choice)) {
 272                                 /* a single choice */
 273                                 $disp = $choice[0];
 274
 275                                 /* fill the carray if the conversion is for the whole article*/
 276                                 if($flag == 'A') {
 277                                         foreach($this->mVariants as $v) {
 278                                                 $carray[$v] = $disp;
 279                                         }
 280                                 }
 281                         }
 282                         else {
 283                                 foreach($choice as $c) {
 284                                         $v = explode($this->mMarkup['codesep'], $c);
 285                                         if(sizeof($v) != 2) // syntax error, skip
 286                                                 continue;
 287                                         $carray[trim($v[0])] = trim($v[1]);
 288                                 }
 289                                 if(array_key_exists($plang, $carray))
 290                                         $disp = $carray[$plang];
 291                                 else if(array_key_exists($fallback, $carray))
 292                                         $disp = $carray[$fallback];
 293                         }
 294                         if(empty($disp)) { // syntax error
 295                                 $text .= $marked[0];
 296                         }
 297                         else {
 298                                 if($flag == 'T') // for title only
 299                                         $this->mTitleDisplay = $disp;
 300                                 else {
 301                                         $text .= $disp;
 302                                         if($flag == 'A') {
 303                                                 /* modify the conversion table for this session*/
 304
 305                                                 /* fill in the missing variants, if any,
 306                                                     with fallbacks */
 307                                                 foreach($this->mVariants as $v) {
 308                                                         if(!array_key_exists($v, $carray)) {
 309                                                                 $vf = $this->getVariantFallback($v);
 310                                                                 if(array_key_exists($vf, $carray))
 311                                                                         $carray[$v] = $carray[$vf];
 312                                                         }
 313                                                 }
 314                                                 foreach($this->mVariants as $vfrom) {
 315                                                         if(!array_key_exists($vfrom, $carray))
 316                                                                 continue;
 317                                                         foreach($this->mVariants as $vto) {
 318                                                                 if($vfrom == $vto)
 319                                                                         continue;
 320                                                                 if(!array_key_exists($vto, $carray))
 321                                                                         continue;
 322                                                                 $this->mTables[$vto][$carray[$vfrom]] = $carray[$vto];
 323                                                         }
 324                                                 }
 325                                         }
 326                                 }
 327                         }
 328                         if(array_key_exists(1, $marked))
 329                                 $text .= $this->autoConvert($marked[1]);
 330                 }
 331
 332                 return $text;
 333         }
 334
 335
 336         /**
 337          * if a language supports multiple variants, it is
 338          * possible that non-existing link in one variant
 339          * actually exists in another variant. this function
 340          * tries to find it. See e.g. LanguageZh.php
 341          *
 342          * @param string $link the name of the link
 343          * @param mixed $nt the title object of the link
 344          * @return null the input parameters may be modified upon return
 345      * @access public
 346          */
 347         function findVariantLink( &$link, &$nt ) {
 348                 static $count=0; //used to limit this operation
 349                 static $cache=array();
 350                 global $wgDisableLangConversion;
 351                 $pref = $this->getPreferredVariant();
 352                 if( $count > 50 )
 353                         return;
 354                 $count++;
 355                 $variants = $this->autoConvertToAllVariants($link);
 356                 if($variants == false) //give up
 357                         return;
 358                 foreach( $variants as $v ) {
 359                         if(isset($cache[$v]))
 360                                 continue;
 361                         $cache[$v] = 1;
 362                         $varnt = Title::newFromText( $v );
 363                         if( $varnt && $varnt->getArticleID() > 0 ) {
 364                                 $nt = $varnt;
 365                                 if( !$wgDisableLangConversion && $pref != 'zh' )
 366                                         $link = $v;
 367                                 break;
 368                         }
 369                 }
 370         }
 371
 372     /**
 373      * returns language specific hash options
 374      *
 375      * @access public
 376      */
 377         function getExtraHashOptions() {
 378                 $variant = $this->getPreferredVariant();
 379                 return '!' . $variant ;
 380         }
 381
 382     /**
 383      * get title text as defined in the body of the article text
 384      *
 385      * @access public
 386      */
 387         function getParsedTitle() {
 388                 return $this->mTitleDisplay;
 389         }
 390
 391         /**
 392      * a write lock to the cache
 393      *
 394      * @access private
 395      */
 396         function lockCache() {
 397                 global $wgMemc;
 398                 $success = false;
 399                 for($i=0; $i<30; $i++) {
 400                         if($success = $wgMemc->add($this->mCacheKey . "lock", 1, 10))
 401                                 break;
 402                         sleep(1);
 403                 }
 404                 return $success;
 405         }
 406
 407         /**
 408      * unlock cache
 409      *
 410      * @access private
 411      */
 412         function unlockCache() {
 413                 global $wgMemc;
 414                 $wgMemc->delete($this->mCacheKey . "lock");
 415         }
 416
 417
 418         /**
 419      * Load default conversion tables
 420      * This method must be implemented in derived class
 421      *
 422      * @access private
 423      */
 424         function loadDefaultTables() {
 425                 $name = get_class($this);
 426                 die("Must implement loadDefaultTables() method in class $name");
 427         }
 428
 429         /**
 430      * load conversion tables either from the cache or the disk
 431      * @access private
 432      */
 433         function loadTables($fromcache=true) {
 434                 global $wgMemc;
 435                 if( $this->mTablesLoaded )
 436                         return;
 437                 $this->mTablesLoaded = true;
 438                 if($fromcache) {
 439                         $this->mTables = $wgMemc->get( $this->mCacheKey );
 440                         if( !empty( $this->mTables ) ) //all done
 441                                 return;
 442                 }
 443                 // not in cache, or we need a fresh reload.
 444                 // we will first load the default tables
 445                 // then update them using things in MediaWiki:Zhconversiontable/*
 446                 global $wgMessageCache;
 447                 $this->loadDefaultTables();
 448                 foreach($this->mVariants as $var) {
 449                         $cached = $this->parseCachedTable($var);
 450                         $this->mTables[$var] = array_merge($this->mTables[$var], $cached);
 451                 }
 452
 453                 $this->postLoadTables();
 454
 455                 if($this->lockCache()) {
 456                         $wgMemc->set($this->mCacheKey, $this->mTables, 43200);
 457                         $this->unlockCache();
 458                 }
 459         }
 460
 461     /**
 462      * Hook for post processig after conversion tables are loaded
 463      *
 464      */
 465         function postLoadTables() {}
 466
 467         /* deprecated? */
 468         function updateTablexxxx($code, $table) {
 469                 global $wgMemc;
 470                 if(!$this->mTablesLoaded)
 471                         $this->loadTables();
 472
 473                 $this->mTables[$code] = array_merge($this->mTables[$code], $table);
 474                 if($this->lockCache()) {
 475                         $wgMemc->delete($this->mCacheKey);
 476                         $wgMemc->set($this->mCacheKey, $this->mTables, 43200);
 477                         $this->unlockCache();
 478                 }
 479         }
 480
 481     /**
 482      * Reload the conversion tables
 483      *
 484      * @access private
 485      */
 486         function reloadTables() {
 487                 if($this->mTables)
 488                         unset($this->mTables);
 489                 $this->mTablesLoaded = false;
 490                 $this->loadTables(false);
 491         }
 492
 493
 494         /**
 495      * parse the conversion table stored in the cache
 496      *
 497      * the tables should be in blocks of the following form:
 498
 499      *          -{
 500      *                  word => word ;
 501      *                  word => word ;
 502      *                  ...
 503      *          }-
 504      *
 505      *  to make the tables more manageable, subpages are allowed
 506      *  and will be parsed recursively if $recursive=true
 507      *
 508      * @access private
 509          */
 510         function parseCachedTable($code, $subpage='', $recursive=true) {
 511                 global $wgMessageCache;
 512                 static $parsed = array();
 513
 514                 if(!is_object($wgMessageCache))
 515                         return array();
 516
 517                 $key = 'Conversiontable/'.$code;
 518                 if($subpage)
 519                         $key .= '/' . $subpage;
 520
 521                 if(array_key_exists($key, $parsed))
 522                         return array();
 523
 524
 525                 $txt = $wgMessageCache->get( $key, true, true, true );
 526
 527                 // get all subpage links of the form
 528                 // [[MediaWiki:conversiontable/zh-xx/...|...]]
 529                 $linkhead = $this->mLangObj->getNsText(NS_MEDIAWIKI) . ':Conversiontable';
 530                 $subs = explode('[[', $txt);
 531                 $sublinks = array();
 532                 foreach( $subs as $sub ) {
 533                         $link = explode(']]', $sub, 2);
 534                         if(count($link) != 2)
 535                                 continue;
 536                         $b = explode('|', $link[0]);
 537                         $b = explode('/', trim($b[0]), 3);
 538                         if(count($b)==3)
 539                                 $sublink = $b[2];
 540                         else
 541                                 $sublink = '';
 542
 543                         if($b[0] == $linkhead && $b[1] == $code) {
 544                                 $sublinks[] = $sublink;
 545                         }
 546                 }
 547
 548
 549                 // parse the mappings in this page
 550                 $blocks = explode('-{', $txt);
 551                 array_shift($blocks);
 552                 $ret = array();
 553                 foreach($blocks as $block) {
 554                         $mappings = explode('}-', $block, 2);
 555                         $stripped = str_replace(array("'", '"', '*','#'), '', $mappings[0]);
 556                         $table = explode( ';', $stripped );
 557                         foreach( $table as $t ) {
 558                                 $m = explode( '=>', $t );
 559                                 if( count( $m ) != 2)
 560                                         continue;
 561                                 // trim any trailling comments starting with '//'
 562                                 $tt = explode('//', $m[1], 2);
 563                                 $ret[trim($m[0])] = trim($tt[0]);
 564                         }
 565                 }
 566                 $parsed[$key] = true;
 567
 568
 569                 // recursively parse the subpages
 570                 if($recursive) {
 571                         foreach($sublinks as $link) {
 572                                 $s = $this->parseCachedTable($code, $link, $recursive);
 573                                 $ret = array_merge($ret, $s);
 574                         }
 575                 }
 576                 return $ret;
 577         }
 578
 579         /**
 580          * Enclose a string with the "no conversion" tag. This is used by
 581          * various functions in the Parser
 582          *
 583          * @param string $text text to be tagged for no conversion
 584          * @return string the tagged text
 585         */
 586         function markNoConversion($text) {
 587                 $ret = $this->mMarkup['begin'] . $text . $this->mMarkup['end'];
 588         }
 589
 590         /**
 591      * hook to refresh the cache of conversion tables when
 592      * MediaWiki:conversiontable* is updated
 593      * @access private
 594         */
 595         function OnArticleSaveComplete($article, $user, $text, $summary, $isminor, $iswatch, $section) {
 596                 $titleobj = $article->getTitle();
 597                 if($titleobj->getNamespace() == NS_MEDIAWIKI) {
 598             /*
 599                         global $wgContLang; // should be an LanguageZh.
 600                         if(get_class($wgContLang) != 'languagezh')
 601                                 return true;
 602             */
 603                         $title = $titleobj->getDBkey();
 604                         $t = explode('/', $title, 3);
 605                         $c = count($t);
 606                         if( $c > 1 && $t[0] == 'Conversiontable' ) {
 607                                 if(in_array($t[1], $this->mVariants)) {
 608                                         $this->reloadTables();
 609                                 }
 610                         }
 611                 }
 612                 return true;
 613         }
 614 }
 615
 616 ?>