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 $mUseFss = false;
  16         var $mTables;
  17         var $mFssObjects;
  18         var $mTitleDisplay='';
  19         var $mDoTitleConvert=true, $mDoContentConvert=true;
  20         var $mCacheKey;
  21         var $mLangObj;
  22         var $mMarkup;
  23         var $mFlags;
  24         var $mUcfirst = false;
  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          * @param array $flags array defining the custom strings that maps to the flags
  33      * @access public
  34      */
  35         function __construct($langobj, $maincode,
  36                                                                 $variants=array(),
  37                                                                 $variantfallbacks=array(),
  38                                                                 $markup=array(),
  39                                                                 $flags = array()) {
  40                 global $wgDBname;
  41                 $this->mLangObj = $langobj;
  42                 $this->mMainLanguageCode = $maincode;
  43                 $this->mVariants = $variants;
  44                 $this->mVariantFallbacks = $variantfallbacks;
  45                 $this->mCacheKey = $wgDBname . ":conversiontables";
  46                 $m = array('begin'=>'-{', 'flagsep'=>'|', 'codesep'=>':',
  47                                    'varsep'=>';', 'end'=>'}-');
  48                 $this->mMarkup = array_merge($m, $markup);
  49                 $f = array('A'=>'A', 'T'=>'T');
  50                 $this->mFlags = array_merge($f, $flags);
  51                 if ( function_exists( 'fss_prep_replace' ) ) {
  52                         $this->mUseFss = true;
  53                 }
  54         }
  55
  56         /**
  57      * @access public
  58      */
  59         function getVariants() {
  60                 return $this->mVariants;
  61         }
  62
  63         /**
  64          * in case some variant is not defined in the markup, we need
  65          * to have some fallback. for example, in zh, normally people
  66          * will define zh-cn and zh-tw, but less so for zh-sg or zh-hk.
  67          * when zh-sg is preferred but not defined, we will pick zh-cn
  68          * in this case. right now this is only used by zh.
  69          *
  70          * @param string $v the language code of the variant
  71          * @return string the code of the fallback language or false if there is no fallback
  72      * @private
  73         */
  74         function getVariantFallback($v) {
  75                 return $this->mVariantFallbacks[$v];
  76         }
  77
  78
  79         /**
  80          * get preferred language variants.
  81          * @param boolean $fromUser Get it from $wgUser's preferences
  82      * @return string the preferred language code
  83      * @access public
  84         */
  85         function getPreferredVariant( $fromUser = true ) {
  86                 global $wgUser, $wgRequest;
  87
  88                 if($this->mPreferredVariant)
  89                         return $this->mPreferredVariant;
  90
  91                 // see if the preference is set in the request
  92                 $req = $wgRequest->getText( 'variant' );
  93                 if( in_array( $req, $this->mVariants ) ) {
  94                         $this->mPreferredVariant = $req;
  95                         return $req;
  96                 }
  97
  98                 // get language variant preference from logged in users
  99                 // Don't call this on stub objects because that causes infinite
 100                 // recursion during initialisation
 101                 if( $fromUser && $wgUser->isLoggedIn() )  {
 102                         $this->mPreferredVariant = $wgUser->getOption('variant');
 103                         return $this->mPreferredVariant;
 104                 }
 105
 106                 # FIXME rewrite code for parsing http header. The current code
 107                 # is written specific for detecting zh- variants
 108                 if( !$this->mPreferredVariant ) {
 109                         // see if some supported language variant is set in the
 110                         // http header, but we don't set the mPreferredVariant
 111                         // variable in case this is called before the user's
 112                         // preference is loaded
 113                         $pv=$this->mMainLanguageCode;
 114                         if(array_key_exists('HTTP_ACCEPT_LANGUAGE', $_SERVER)) {
 115                                 $header = str_replace( '_', '-', strtolower($_SERVER["HTTP_ACCEPT_LANGUAGE"]));
 116                                 $zh = strstr($header, 'zh-');
 117                                 if($zh) {
 118                                         $pv = substr($zh,0,5);
 119                                 }
 120                         }
 121                         return $pv;
 122                 }
 123         }
 124
 125         /**
 126      * dictionary-based conversion
 127      *
 128      * @param string $text the text to be converted
 129      * @param string $toVariant the target language code
 130      * @return string the converted text
 131      * @private
 132      */
 133         function autoConvert($text, $toVariant=false) {
 134                 $fname="LanguageConverter::autoConvert";
 135
 136                 wfProfileIn( $fname );
 137
 138                 if(!$this->mTablesLoaded)
 139                         $this->loadTables();
 140
 141                 if(!$toVariant)
 142                         $toVariant = $this->getPreferredVariant();
 143                 if(!in_array($toVariant, $this->mVariants))
 144                         return $text;
 145
 146                 /* we convert everything except:
 147                    1. html markups (anything between < and >)
 148                    2. html entities
 149                    3. place holders created by the parser
 150                 */
 151                 global $wgParser;
 152                 if (isset($wgParser))
 153                         $marker = '|' . $wgParser->UniqPrefix() . '[\-a-zA-Z0-9]+';
 154                 else
 155                         $marker = "";
 156
 157                 // this one is needed when the text is inside an html markup
 158                 $htmlfix = '|<[^>]+=\"[^(>=)]*$|^[^(<>=\")]*\"[^>]*>';
 159
 160                 $reg = '/<[^>]+>|&[a-z#][a-z0-9]+;' . $marker . $htmlfix . '/';
 161
 162                 $matches = preg_split($reg, $text, -1, PREG_SPLIT_OFFSET_CAPTURE);
 163
 164
 165                 $m = array_shift($matches);
 166                 $ret = $this->translate($m[0], $toVariant);
 167                 $mstart = $m[1]+strlen($m[0]);
 168                 foreach($matches as $m) {
 169                         $ret .= substr($text, $mstart, $m[1]-$mstart);
 170                         $ret .= $this->translate($m[0], $toVariant);
 171                         $mstart = $m[1] + strlen($m[0]);
 172                 }
 173                 wfProfileOut( $fname );
 174                 return $ret;
 175         }
 176
 177         /**
 178          * Translate a string to a variant
 179          * Doesn't process markup or do any of that other stuff, for that use convert()
 180          *
 181          * @param string $text Text to convert
 182          * @param string $variant Variant language code
 183          * @return string Translated text
 184          */
 185         function translate( $text, $variant ) {
 186                 if( !$this->mTablesLoaded )
 187                         $this->loadTables();
 188                 if ( $this->mUseFss ) {
 189                         return fss_exec_replace( $this->mFssObjects[$variant], $text );
 190                 } else {
 191                         return strtr( $text, $this->mTables[$variant] );
 192                 }
 193         }
 194
 195         /**
 196      * convert text to all supported variants
 197      *
 198      * @param string $text the text to be converted
 199      * @return array of string
 200      * @private
 201      */
 202         function autoConvertToAllVariants($text) {
 203                 $fname="LanguageConverter::autoConvertToAllVariants";
 204                 wfProfileIn( $fname );
 205                 if( !$this->mTablesLoaded )
 206                         $this->loadTables();
 207
 208                 $ret = array();
 209                 foreach($this->mVariants as $variant) {
 210                         $ret[$variant] = $this->translate($text, $variant);
 211                 }
 212                 wfProfileOut( $fname );
 213                 return $ret;
 214         }
 215
 216         /**
 217          * Convert text using a parser object for context
 218          */
 219         function parserConvert( $text, &$parser ) {
 220                 global $wgDisableLangConversion;
 221                 /* don't do anything if this is the conversion table */
 222                 if ( $parser->mTitle->getNamespace() == NS_MEDIAWIKI &&
 223                         strpos($parser->mTitle->getText, "Conversiontable") !== false )
 224                 {
 225                         return $text;
 226                 }
 227
 228                 if($wgDisableLangConversion)
 229                         return $text;
 230
 231                 $text = $this->convert( $text );
 232                 $parser->mOutput->setTitleText( $this->mTitleDisplay );
 233                 return $text;
 234         }
 235
 236         /**
 237          * convert text to different variants of a language. the automatic
 238          * conversion is done in autoConvert(). here we parse the text
 239          * marked with -{}-, which specifies special conversions of the
 240          * text that can not be accomplished in autoConvert()
 241          *
 242          * syntax of the markup:
 243          * -{code1:text1;code2:text2;...}-  or
 244          * -{text}- in which case no conversion should take place for text
 245      *
 246      * @param string $text text to be converted
 247      * @param bool $isTitle whether this conversion is for the article title
 248      * @return string converted text
 249      * @access public
 250      */
 251         function convert( $text , $isTitle=false) {
 252                 $mw =& MagicWord::get( 'notitleconvert'   );
 253                 if( $mw->matchAndRemove( $text ) )
 254                         $this->mDoTitleConvert = false;
 255
 256                 $mw =& MagicWord::get( 'nocontentconvert'   );
 257                 if( $mw->matchAndRemove( $text ) ) {
 258                         $this->mDoContentConvert = false;
 259                 }
 260
 261                 // no conversion if redirecting
 262                 $mw =& MagicWord::get( 'redirect'   );
 263                 if( $mw->matchStart( $text ))
 264                         return $text;
 265
 266                 if( $isTitle ) {
 267                         if( !$this->mDoTitleConvert ) {
 268                                 $this->mTitleDisplay = $text;
 269                                 return $text;
 270                         }
 271                         if( !empty($this->mTitleDisplay))
 272                                 return $this->mTitleDisplay;
 273
 274                         global $wgRequest;
 275                         $isredir = $wgRequest->getText( 'redirect', 'yes' );
 276                         $action = $wgRequest->getText( 'action' );
 277                         if ( $isredir == 'no' || $action == 'edit' ) {
 278                                 return $text;
 279                         }
 280                         else {
 281                                 $this->mTitleDisplay = $this->autoConvert($text);
 282                                 return $this->mTitleDisplay;
 283                         }
 284                 }
 285
 286                 if( !$this->mDoContentConvert )
 287                         return $text;
 288
 289                 $plang = $this->getPreferredVariant();
 290                 if( isset( $this->mVariantFallbacks[$plang] ) ) {
 291                         $fallback = $this->mVariantFallbacks[$plang];
 292                 } else {
 293                         // This sounds... bad?
 294                         $fallback = '';
 295                 }
 296
 297                 $tarray = explode($this->mMarkup['begin'], $text);
 298                 $tfirst = array_shift($tarray);
 299                 $text = $this->autoConvert($tfirst);
 300                 foreach($tarray as $txt) {
 301                         $marked = explode($this->mMarkup['end'], $txt, 2);
 302                         $flags = array();
 303                         $tt = explode($this->mMarkup['flagsep'], $marked[0], 2);
 304
 305                         if(sizeof($tt) == 2) {
 306                                 $f = explode($this->mMarkup['varsep'], $tt[0]);
 307                                 foreach($f as $ff) {
 308                                         $ff = trim($ff);
 309                                         if(array_key_exists($ff, $this->mFlags) &&
 310                                                 !array_key_exists($this->mFlags[$ff], $flags))
 311                                                 $flags[] = $this->mFlags[$ff];
 312                                 }
 313                                 $rules = $tt[1];
 314                         }
 315                         else
 316                                 $rules = $marked[0];
 317
 318 #FIXME: may cause trouble here...
 319                         //strip &nbsp; since it interferes with the parsing, plus,
 320                         //all spaces should be stripped in this tag anyway.
 321                         $rules = str_replace('&nbsp;', '', $rules);
 322
 323                         $carray = $this->parseManualRule($rules, $flags);
 324                         $disp = '';
 325                         if(array_key_exists($plang, $carray))
 326                                 $disp = $carray[$plang];
 327                         else if(array_key_exists($fallback, $carray))
 328                                 $disp = $carray[$fallback];
 329                         if($disp) {
 330                                 if(in_array('T',  $flags))
 331                                         $this->mTitleDisplay = $disp;
 332                                 else
 333                                         $text .= $disp;
 334
 335                                 if(in_array('A', $flags)) {
 336                                         /* modify the conversion table for this session*/
 337
 338                                         /* fill in the missing variants, if any,
 339                                             with fallbacks */
 340                                         foreach($this->mVariants as $v) {
 341                                                 if(!array_key_exists($v, $carray)) {
 342                                                         $vf = $this->getVariantFallback($v);
 343                                                         if(array_key_exists($vf, $carray))
 344                                                                 $carray[$v] = $carray[$vf];
 345                                                 }
 346                                         }
 347
 348                                         foreach($this->mVariants as $vfrom) {
 349                                                 if(!array_key_exists($vfrom, $carray))
 350                                                         continue;
 351                                                 foreach($this->mVariants as $vto) {
 352                                                         if($vfrom == $vto)
 353                                                                 continue;
 354                                                         if(!array_key_exists($vto, $carray))
 355                                                                 continue;
 356                                                         $this->mTables[$vto][$carray[$vfrom]] = $carray[$vto];
 357
 358                                                 }
 359                                         }
 360                                         if ( $this->mUseFss ) {
 361                                                 $this->generateFssObjects();
 362                                         }
 363                                 }
 364                         }
 365                         else {
 366                                 $text .= $marked[0];
 367                         }
 368                         if(array_key_exists(1, $marked))
 369                                 $text .= $this->autoConvert($marked[1]);
 370                 }
 371
 372                 return $text;
 373         }
 374
 375         /**
 376          * parse the manually marked conversion rule
 377          * @param string $rule the text of the rule
 378          * @return array of the translation in each variant
 379          * @private
 380          */
 381         function parseManualRule($rules, $flags=array()) {
 382
 383                 $choice = explode($this->mMarkup['varsep'], $rules);
 384                 $carray = array();
 385                 if(sizeof($choice) == 1) {
 386                         /* a single choice */
 387                         foreach($this->mVariants as $v)
 388                                 $carray[$v] = $choice[0];
 389                 }
 390                 else {
 391                         foreach($choice as $c) {
 392                                 $v = explode($this->mMarkup['codesep'], $c);
 393                                 if(sizeof($v) != 2) // syntax error, skip
 394                                         continue;
 395                                 $carray[trim($v[0])] = trim($v[1]);
 396                         }
 397                 }
 398                 return $carray;
 399         }
 400
 401         /**
 402          * if a language supports multiple variants, it is
 403          * possible that non-existing link in one variant
 404          * actually exists in another variant. this function
 405          * tries to find it. See e.g. LanguageZh.php
 406          *
 407          * @param string $link the name of the link
 408          * @param mixed $nt the title object of the link
 409          * @return null the input parameters may be modified upon return
 410      * @access public
 411          */
 412         function findVariantLink( &$link, &$nt ) {
 413                 static $count=0; //used to limit this operation
 414                 static $cache=array();
 415                 global $wgDisableLangConversion;
 416                 $pref = $this->getPreferredVariant();
 417                 $ns=0;
 418                 if(is_object($nt))
 419                         $ns = $nt->getNamespace();
 420                 if( $count > 50 && $ns != NS_CATEGORY )
 421                         return;
 422                 $count++;
 423                 $variants = $this->autoConvertToAllVariants($link);
 424                 if($variants == false) //give up
 425                         return;
 426                 foreach( $variants as $v ) {
 427                         if(isset($cache[$v]))
 428                                 continue;
 429                         $cache[$v] = 1;
 430                         $varnt = Title::newFromText( $v, $ns );
 431                         if( $varnt && $varnt->getArticleID() > 0 ) {
 432                                 $nt = $varnt;
 433                                 if( !$wgDisableLangConversion )
 434                                         $link = $v;
 435                                 break;
 436                         }
 437                 }
 438         }
 439
 440     /**
 441      * returns language specific hash options
 442      *
 443      * @access public
 444      */
 445         function getExtraHashOptions() {
 446                 $variant = $this->getPreferredVariant();
 447                 return '!' . $variant ;
 448         }
 449
 450     /**
 451      * get title text as defined in the body of the article text
 452      *
 453      * @access public
 454      */
 455         function getParsedTitle() {
 456                 return $this->mTitleDisplay;
 457         }
 458
 459         /**
 460      * a write lock to the cache
 461      *
 462      * @private
 463      */
 464         function lockCache() {
 465                 global $wgMemc;
 466                 $success = false;
 467                 for($i=0; $i<30; $i++) {
 468                         if($success = $wgMemc->add($this->mCacheKey . "lock", 1, 10))
 469                                 break;
 470                         sleep(1);
 471                 }
 472                 return $success;
 473         }
 474
 475         /**
 476      * unlock cache
 477      *
 478      * @private
 479      */
 480         function unlockCache() {
 481                 global $wgMemc;
 482                 $wgMemc->delete($this->mCacheKey . "lock");
 483         }
 484
 485
 486         /**
 487      * Load default conversion tables
 488      * This method must be implemented in derived class
 489      *
 490      * @private
 491      */
 492         function loadDefaultTables() {
 493                 $name = get_class($this);
 494                 wfDie("Must implement loadDefaultTables() method in class $name");
 495         }
 496
 497         /**
 498      * load conversion tables either from the cache or the disk
 499      * @private
 500      */
 501         function loadTables($fromcache=true) {
 502                 global $wgMemc;
 503                 if( $this->mTablesLoaded )
 504                         return;
 505                 $this->mTablesLoaded = true;
 506                 if($fromcache) {
 507                         $this->mTables = $wgMemc->get( $this->mCacheKey );
 508                         if( !empty( $this->mTables ) ) //all done
 509                                 return;
 510                 }
 511                 // not in cache, or we need a fresh reload.
 512                 // we will first load the default tables
 513                 // then update them using things in MediaWiki:Zhconversiontable/*
 514                 global $wgMessageCache;
 515                 $this->loadDefaultTables();
 516                 foreach($this->mVariants as $var) {
 517                         $cached = $this->parseCachedTable($var);
 518                         $this->mTables[$var] = array_merge($this->mTables[$var], $cached);
 519                 }
 520
 521                 $this->postLoadTables();
 522
 523                 if($this->lockCache()) {
 524                         $wgMemc->set($this->mCacheKey, $this->mTables, 43200);
 525                         $this->unlockCache();
 526                 }
 527                 if ( $this->mUseFss ) {
 528                         $this->generateFssObjects();
 529                 }
 530         }
 531
 532         /**
 533          * Generate FSS objects. The FSS extension must be available.
 534          */
 535         function generateFssObjects() {
 536                 foreach ( $this->mTables as $variant => $table ) {
 537                         $this->mFssObjects[$variant] = fss_prep_replace( $table );
 538                 }
 539         }
 540
 541     /**
 542      * Hook for post processig after conversion tables are loaded
 543      *
 544      */
 545         function postLoadTables() {}
 546
 547     /**
 548      * Reload the conversion tables
 549      *
 550      * @private
 551      */
 552         function reloadTables() {
 553                 if($this->mTables)
 554                         unset($this->mTables);
 555                 $this->mTablesLoaded = false;
 556                 $this->loadTables(false);
 557         }
 558
 559
 560         /**
 561      * parse the conversion table stored in the cache
 562      *
 563      * the tables should be in blocks of the following form:
 564
 565      *          -{
 566      *                  word => word ;
 567      *                  word => word ;
 568      *                  ...
 569      *          }-
 570      *
 571      *  to make the tables more manageable, subpages are allowed
 572      *  and will be parsed recursively if $recursive=true
 573      *
 574      * @private
 575          */
 576         function parseCachedTable($code, $subpage='', $recursive=true) {
 577                 global $wgMessageCache;
 578                 static $parsed = array();
 579
 580                 if(!is_object($wgMessageCache))
 581                         return array();
 582
 583                 $key = 'Conversiontable/'.$code;
 584                 if($subpage)
 585                         $key .= '/' . $subpage;
 586
 587                 if(array_key_exists($key, $parsed))
 588                         return array();
 589
 590
 591                 $txt = $wgMessageCache->get( $key, true, true, true );
 592
 593                 // get all subpage links of the form
 594                 // [[MediaWiki:conversiontable/zh-xx/...|...]]
 595                 $linkhead = $this->mLangObj->getNsText(NS_MEDIAWIKI) . ':Conversiontable';
 596                 $subs = explode('[[', $txt);
 597                 $sublinks = array();
 598                 foreach( $subs as $sub ) {
 599                         $link = explode(']]', $sub, 2);
 600                         if(count($link) != 2)
 601                                 continue;
 602                         $b = explode('|', $link[0]);
 603                         $b = explode('/', trim($b[0]), 3);
 604                         if(count($b)==3)
 605                                 $sublink = $b[2];
 606                         else
 607                                 $sublink = '';
 608
 609                         if($b[0] == $linkhead && $b[1] == $code) {
 610                                 $sublinks[] = $sublink;
 611                         }
 612                 }
 613
 614
 615                 // parse the mappings in this page
 616                 $blocks = explode($this->mMarkup['begin'], $txt);
 617                 array_shift($blocks);
 618                 $ret = array();
 619                 foreach($blocks as $block) {
 620                         $mappings = explode($this->mMarkup['end'], $block, 2);
 621                         $stripped = str_replace(array("'", '"', '*','#'), '', $mappings[0]);
 622                         $table = explode( ';', $stripped );
 623                         foreach( $table as $t ) {
 624                                 $m = explode( '=>', $t );
 625                                 if( count( $m ) != 2)
 626                                         continue;
 627                                 // trim any trailling comments starting with '//'
 628                                 $tt = explode('//', $m[1], 2);
 629                                 $ret[trim($m[0])] = trim($tt[0]);
 630                         }
 631                 }
 632                 $parsed[$key] = true;
 633
 634
 635                 // recursively parse the subpages
 636                 if($recursive) {
 637                         foreach($sublinks as $link) {
 638                                 $s = $this->parseCachedTable($code, $link, $recursive);
 639                                 $ret = array_merge($ret, $s);
 640                         }
 641                 }
 642
 643                 if ($this->mUcfirst) {
 644                         foreach ($ret as $k => $v) {
 645                                 $ret[Language::ucfirst($k)] = Language::ucfirst($v);
 646                         }
 647                 }
 648                 return $ret;
 649         }
 650
 651         /**
 652          * Enclose a string with the "no conversion" tag. This is used by
 653          * various functions in the Parser
 654          *
 655          * @param string $text text to be tagged for no conversion
 656          * @return string the tagged text
 657         */
 658         function markNoConversion($text) {
 659                 # don't mark if already marked
 660                 if(strpos($text, $this->mMarkup['begin']) ||
 661                    strpos($text, $this->mMarkup['end']))
 662                         return $text;
 663
 664                 $ret = $this->mMarkup['begin'] . $text . $this->mMarkup['end'];
 665                 return $ret;
 666         }
 667
 668         /**
 669          * convert the sorting key for category links. this should make different
 670          * keys that are variants of each other map to the same key
 671         */
 672         function convertCategoryKey( $key ) {
 673                 return $key;
 674         }
 675         /**
 676      * hook to refresh the cache of conversion tables when
 677      * MediaWiki:conversiontable* is updated
 678      * @private
 679         */
 680         function OnArticleSaveComplete($article, $user, $text, $summary, $isminor, $iswatch, $section) {
 681                 $titleobj = $article->getTitle();
 682                 if($titleobj->getNamespace() == NS_MEDIAWIKI) {
 683             /*
 684                         global $wgContLang; // should be an LanguageZh.
 685                         if(get_class($wgContLang) != 'languagezh')
 686                                 return true;
 687             */
 688                         $title = $titleobj->getDBkey();
 689                         $t = explode('/', $title, 3);
 690                         $c = count($t);
 691                         if( $c > 1 && $t[0] == 'Conversiontable' ) {
 692                                 if(in_array($t[1], $this->mVariants)) {
 693                                         $this->reloadTables();
 694                                 }
 695                         }
 696                 }
 697                 return true;
 698         }
 699 }
 700
 701 ?>