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         var $mFlags;
  22         /**
  23      * Constructor
  24          *
  25      * @param string $maincode the main language code of this language
  26      * @param array $variants the supported variants of this language
  27      * @param array $variantfallback the fallback language of each variant
  28      * @param array $markup array defining the markup used for manual conversion
  29          * @param array $flags array defining the custom strings that maps to the flags
  30      * @access public
  31      */
  32         function LanguageConverter($langobj, $maincode,
  33                                                                 $variants=array(),
  34                                                                 $variantfallbacks=array(),
  35                                                                 $markup=array(),
  36                                                                 $flags = array()) {
  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                 $m = array('begin'=>'-{', 'flagsep'=>'|', 'codesep'=>':',
  44                                    'varsep'=>';', 'end'=>'}-');
  45                 $this->mMarkup = array_merge($m, $markup);
  46                 $f = array('A'=>'A', 'T'=>'T');
  47                 $this->mFlags = array_merge($f, $flags);
  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
 124                 wfProfileIn( $fname );
 125
 126                 if(!$this->mTablesLoaded)
 127                         $this->loadTables();
 128
 129                 if(!$toVariant)
 130                         $toVariant = $this->getPreferredVariant();
 131                 if(!in_array($toVariant, $this->mVariants))
 132                         return $text;
 133
 134
 135                 $reg = '/<[^>]+>|&[a-z#][a-z0-9]+;|'.UNIQ_PREFIX.'-[a-zA-Z0-9]+/';
 136                 $matches = preg_split($reg, $text, -1, PREG_SPLIT_OFFSET_CAPTURE);
 137
 138
 139                 $m = array_shift($matches);
 140                 $ret = strtr($m[0], $this->mTables[$toVariant]);
 141                 $mstart = $m[1]+strlen($m[0]);
 142                 foreach($matches as $m) {
 143                         $ret .= substr($text, $mstart, $m[1]-$mstart);
 144                         $ret .= strtr($m[0], $this->mTables[$toVariant]);
 145                         $mstart = $m[1] + strlen($m[0]);
 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                 global $wgTitle;
 190
 191                 /* don't do anything if this is the conversion table */
 192                 if($wgTitle->getNamespace() == NS_MEDIAWIKI &&
 193                    strpos($wgTitle->getText(), "Conversiontable")!==false)
 194                         return $text;
 195
 196                 if($wgDisableLangConversion)
 197                         return $text;
 198
 199                 $mw =& MagicWord::get( MAG_NOTITLECONVERT );
 200                 if( $mw->matchAndRemove( $text ) )
 201                         $this->mDoTitleConvert = false;
 202
 203                 $mw =& MagicWord::get( MAG_NOCONTENTCONVERT );
 204                 if( $mw->matchAndRemove( $text ) ) {
 205                         $this->mDoContentConvert = false;
 206                 }
 207
 208                 // no conversion if redirecting
 209                 $mw =& MagicWord::get( MAG_REDIRECT );
 210                 if( $mw->matchStart( $text ))
 211                         return $text;
 212
 213                 if( $isTitle ) {
 214                         if( !$this->mDoTitleConvert ) {
 215                                 $this->mTitleDisplay = $text;
 216                                 return $text;
 217                         }
 218                         if( !empty($this->mTitleDisplay))
 219                                 return $this->mTitleDisplay;
 220
 221                         global $wgRequest;
 222                         $isredir = $wgRequest->getText( 'redirect', 'yes' );
 223                         $action = $wgRequest->getText( 'action' );
 224                         if ( $isredir == 'no' || $action == 'edit' ) {
 225                                 return $text;
 226                         }
 227                         else {
 228                                 $this->mTitleDisplay = $this->autoConvert($text);
 229                                 return $this->mTitleDisplay;
 230                         }
 231                 }
 232
 233                 if( !$this->mDoContentConvert )
 234                         return $text;
 235
 236                 $plang = $this->getPreferredVariant();
 237                 $fallback = $this->mVariantFallbacks[$plang];
 238
 239                 $tarray = explode($this->mMarkup['begin'], $text);
 240                 $tfirst = array_shift($tarray);
 241                 $text = $this->autoConvert($tfirst);
 242                 foreach($tarray as $txt) {
 243                         $marked = explode($this->mMarkup['end'], $txt);
 244                         $flags = array();
 245                         $tt = explode($this->mMarkup['flagsep'], $marked[0], 2);
 246
 247                         if(sizeof($tt) == 2) {
 248                                 $f = explode($this->mMarkup['varsep'], $tt[0]);
 249                                 foreach($f as $ff) {
 250                                         $ff = trim($ff);
 251                                         if(array_key_exists($ff, $this->mFlags) &&
 252                                                 !array_key_exists($this->mFlags[$ff], $flags))
 253                                                 $flags[] = $this->mFlags[$ff];
 254                                 }
 255                                 $rules = $tt[1];
 256                         }
 257                         else
 258                                 $rules = $marked[0];
 259
 260 #FIXME: may cause trouble here...
 261                         //strip &nbsp; since it interferes with the parsing, plus,
 262                         //all spaces should be stripped in this tag anyway.
 263                         $rules = str_replace('&nbsp;', '', $rules);
 264
 265                         $carray = $this->parseManualRule($rules, $flags);
 266                         $disp = '';
 267                         if(array_key_exists($plang, $carray))
 268                                 $disp = $carray[$plang];
 269                         else if(array_key_exists($fallback, $carray))
 270                                 $disp = $carray[$fallback];
 271                         if($disp) {
 272                                 if(in_array('T',  $flags))
 273                                         $this->mTitleDisplay = $disp;
 274                                 else
 275                                         $text .= $disp;
 276
 277                                 if(in_array('A', $flags)) {
 278                                         /* modify the conversion table for this session*/
 279
 280                                         /* fill in the missing variants, if any,
 281                                             with fallbacks */
 282                                         foreach($this->mVariants as $v) {
 283                                                 if(!array_key_exists($v, $carray)) {
 284                                                         $vf = $this->getVariantFallback($v);
 285                                                         if(array_key_exists($vf, $carray))
 286                                                                 $carray[$v] = $carray[$vf];
 287                                                 }
 288                                         }
 289
 290                                         foreach($this->mVariants as $vfrom) {
 291                                                 if(!array_key_exists($vfrom, $carray))
 292                                                         continue;
 293                                                 foreach($this->mVariants as $vto) {
 294                                                         if($vfrom == $vto)
 295                                                                 continue;
 296                                                         if(!array_key_exists($vto, $carray))
 297                                                                 continue;
 298                                                         $this->mTables[$vto][$carray[$vfrom]] = $carray[$vto];
 299
 300                                                 }
 301                                         }
 302                                 }
 303                         }
 304                         else {
 305                                 $text .= $marked[0];
 306                         }
 307                         if(array_key_exists(1, $marked))
 308                                 $text .= $this->autoConvert($marked[1]);
 309                 }
 310
 311                 return $text;
 312         }
 313
 314         /**
 315          * parse the manually marked conversion rule
 316          * @param string $rule the text of the rule
 317          * @return array of the translation in each variant
 318          * @access private
 319          */
 320         function parseManualRule($rules, $flags=array()) {
 321
 322                 $choice = explode($this->mMarkup['varsep'], $rules);
 323                 $carray = array();
 324                 if(sizeof($choice) == 1) {
 325                         /* a single choice */
 326                         foreach($this->mVariants as $v)
 327                                 $carray[$v] = $choice[0];
 328                 }
 329                 else {
 330                         foreach($choice as $c) {
 331                                 $v = explode($this->mMarkup['codesep'], $c);
 332                                 if(sizeof($v) != 2) // syntax error, skip
 333                                         continue;
 334                                 $carray[trim($v[0])] = trim($v[1]);
 335                         }
 336                 }
 337                 return $carray;
 338         }
 339
 340         /**
 341          * if a language supports multiple variants, it is
 342          * possible that non-existing link in one variant
 343          * actually exists in another variant. this function
 344          * tries to find it. See e.g. LanguageZh.php
 345          *
 346          * @param string $link the name of the link
 347          * @param mixed $nt the title object of the link
 348          * @return null the input parameters may be modified upon return
 349      * @access public
 350          */
 351         function findVariantLink( &$link, &$nt ) {
 352                 static $count=0; //used to limit this operation
 353                 static $cache=array();
 354                 global $wgDisableLangConversion;
 355                 $pref = $this->getPreferredVariant();
 356                 $ns=0;
 357                 if(is_object($nt))
 358                         $ns = $nt->getNamespace();
 359                 if( $count > 50 && $ns != NS_CATEGORY )
 360                         return;
 361                 $count++;
 362                 $variants = $this->autoConvertToAllVariants($link);
 363                 if($variants == false) //give up
 364                         return;
 365                 foreach( $variants as $v ) {
 366                         if(isset($cache[$v]))
 367                                 continue;
 368                         $cache[$v] = 1;
 369                         $varnt = Title::newFromText( $v );
 370                         if( $varnt && $varnt->getArticleID() > 0 ) {
 371                                 $nt = $varnt;
 372                                 if( !$wgDisableLangConversion )
 373                                         $link = $v;
 374                                 break;
 375                         }
 376                 }
 377         }
 378
 379     /**
 380      * returns language specific hash options
 381      *
 382      * @access public
 383      */
 384         function getExtraHashOptions() {
 385                 $variant = $this->getPreferredVariant();
 386                 return '!' . $variant ;
 387         }
 388
 389     /**
 390      * get title text as defined in the body of the article text
 391      *
 392      * @access public
 393      */
 394         function getParsedTitle() {
 395                 return $this->mTitleDisplay;
 396         }
 397
 398         /**
 399      * a write lock to the cache
 400      *
 401      * @access private
 402      */
 403         function lockCache() {
 404                 global $wgMemc;
 405                 $success = false;
 406                 for($i=0; $i<30; $i++) {
 407                         if($success = $wgMemc->add($this->mCacheKey . "lock", 1, 10))
 408                                 break;
 409                         sleep(1);
 410                 }
 411                 return $success;
 412         }
 413
 414         /**
 415      * unlock cache
 416      *
 417      * @access private
 418      */
 419         function unlockCache() {
 420                 global $wgMemc;
 421                 $wgMemc->delete($this->mCacheKey . "lock");
 422         }
 423
 424
 425         /**
 426      * Load default conversion tables
 427      * This method must be implemented in derived class
 428      *
 429      * @access private
 430      */
 431         function loadDefaultTables() {
 432                 $name = get_class($this);
 433                 die("Must implement loadDefaultTables() method in class $name");
 434         }
 435
 436         /**
 437      * load conversion tables either from the cache or the disk
 438      * @access private
 439      */
 440         function loadTables($fromcache=true) {
 441                 global $wgMemc;
 442                 if( $this->mTablesLoaded )
 443                         return;
 444                 $this->mTablesLoaded = true;
 445                 if($fromcache) {
 446                         $this->mTables = $wgMemc->get( $this->mCacheKey );
 447                         if( !empty( $this->mTables ) ) //all done
 448                                 return;
 449                 }
 450                 // not in cache, or we need a fresh reload.
 451                 // we will first load the default tables
 452                 // then update them using things in MediaWiki:Zhconversiontable/*
 453                 global $wgMessageCache;
 454                 $this->loadDefaultTables();
 455                 foreach($this->mVariants as $var) {
 456                         $cached = $this->parseCachedTable($var);
 457                         $this->mTables[$var] = array_merge($this->mTables[$var], $cached);
 458                 }
 459
 460                 $this->postLoadTables();
 461
 462                 if($this->lockCache()) {
 463                         $wgMemc->set($this->mCacheKey, $this->mTables, 43200);
 464                         $this->unlockCache();
 465                 }
 466         }
 467
 468     /**
 469      * Hook for post processig after conversion tables are loaded
 470      *
 471      */
 472         function postLoadTables() {}
 473
 474     /**
 475      * Reload the conversion tables
 476      *
 477      * @access private
 478      */
 479         function reloadTables() {
 480                 if($this->mTables)
 481                         unset($this->mTables);
 482                 $this->mTablesLoaded = false;
 483                 $this->loadTables(false);
 484         }
 485
 486
 487         /**
 488      * parse the conversion table stored in the cache
 489      *
 490      * the tables should be in blocks of the following form:
 491
 492      *          -{
 493      *                  word => word ;
 494      *                  word => word ;
 495      *                  ...
 496      *          }-
 497      *
 498      *  to make the tables more manageable, subpages are allowed
 499      *  and will be parsed recursively if $recursive=true
 500      *
 501      * @access private
 502          */
 503         function parseCachedTable($code, $subpage='', $recursive=true) {
 504                 global $wgMessageCache;
 505                 static $parsed = array();
 506
 507                 if(!is_object($wgMessageCache))
 508                         return array();
 509
 510                 $key = 'Conversiontable/'.$code;
 511                 if($subpage)
 512                         $key .= '/' . $subpage;
 513
 514                 if(array_key_exists($key, $parsed))
 515                         return array();
 516
 517
 518                 $txt = $wgMessageCache->get( $key, true, true, true );
 519
 520                 // get all subpage links of the form
 521                 // [[MediaWiki:conversiontable/zh-xx/...|...]]
 522                 $linkhead = $this->mLangObj->getNsText(NS_MEDIAWIKI) . ':Conversiontable';
 523                 $subs = explode('[[', $txt);
 524                 $sublinks = array();
 525                 foreach( $subs as $sub ) {
 526                         $link = explode(']]', $sub, 2);
 527                         if(count($link) != 2)
 528                                 continue;
 529                         $b = explode('|', $link[0]);
 530                         $b = explode('/', trim($b[0]), 3);
 531                         if(count($b)==3)
 532                                 $sublink = $b[2];
 533                         else
 534                                 $sublink = '';
 535
 536                         if($b[0] == $linkhead && $b[1] == $code) {
 537                                 $sublinks[] = $sublink;
 538                         }
 539                 }
 540
 541
 542                 // parse the mappings in this page
 543                 $blocks = explode($this->mMarkup['begin'], $txt);
 544                 array_shift($blocks);
 545                 $ret = array();
 546                 foreach($blocks as $block) {
 547                         $mappings = explode($this->mMarkup['end'], $block, 2);
 548                         $stripped = str_replace(array("'", '"', '*','#'), '', $mappings[0]);
 549                         $table = explode( ';', $stripped );
 550                         foreach( $table as $t ) {
 551                                 $m = explode( '=>', $t );
 552                                 if( count( $m ) != 2)
 553                                         continue;
 554                                 // trim any trailling comments starting with '//'
 555                                 $tt = explode('//', $m[1], 2);
 556                                 $ret[trim($m[0])] = trim($tt[0]);
 557                         }
 558                 }
 559                 $parsed[$key] = true;
 560
 561
 562                 // recursively parse the subpages
 563                 if($recursive) {
 564                         foreach($sublinks as $link) {
 565                                 $s = $this->parseCachedTable($code, $link, $recursive);
 566                                 $ret = array_merge($ret, $s);
 567                         }
 568                 }
 569                 return $ret;
 570         }
 571
 572         /**
 573          * Enclose a string with the "no conversion" tag. This is used by
 574          * various functions in the Parser
 575          *
 576          * @param string $text text to be tagged for no conversion
 577          * @return string the tagged text
 578         */
 579         function markNoConversion($text) {
 580                 # don't mark if already marked
 581                 if(strpos($text, $this->mMarkup['begin']) ||
 582                    strpos($text, $this->mMarkup['end']))
 583                         return $text;
 584
 585                 $ret = $this->mMarkup['begin'] . $text . $this->mMarkup['end'];
 586                 return $ret;
 587         }
 588
 589         /**
 590          * convert the sorting key for category links. this should make different
 591          * keys that are variants of each other map to the same key
 592         */
 593         function convertCategoryKey( $key ) {
 594                 return $key;
 595         }
 596         /**
 597      * hook to refresh the cache of conversion tables when
 598      * MediaWiki:conversiontable* is updated
 599      * @access private
 600         */
 601         function OnArticleSaveComplete($article, $user, $text, $summary, $isminor, $iswatch, $section) {
 602                 $titleobj = $article->getTitle();
 603                 if($titleobj->getNamespace() == NS_MEDIAWIKI) {
 604             /*
 605                         global $wgContLang; // should be an LanguageZh.
 606                         if(get_class($wgContLang) != 'languagezh')
 607                                 return true;
 608             */
 609                         $title = $titleobj->getDBkey();
 610                         $t = explode('/', $title, 3);
 611                         $c = count($t);
 612                         if( $c > 1 && $t[0] == 'Conversiontable' ) {
 613                                 if(in_array($t[1], $this->mVariants)) {
 614                                         $this->reloadTables();
 615                                 }
 616                         }
 617                 }
 618                 return true;
 619         }
 620 }
 621
 622 ?>