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         var $mUcfirst = false;
  23         /**
  24      * Constructor
  25          *
  26      * @param string $maincode the main language code of this language
  27      * @param array $variants the supported variants of this language
  28      * @param array $variantfallback the fallback language of each variant
  29      * @param array $markup array defining the markup used for manual conversion
  30          * @param array $flags array defining the custom strings that maps to the flags
  31      * @access public
  32      */
  33         function __construct($langobj, $maincode,
  34                                                                 $variants=array(),
  35                                                                 $variantfallbacks=array(),
  36                                                                 $markup=array(),
  37                                                                 $flags = array()) {
  38                 global $wgDBname;
  39                 $this->mLangObj = $langobj;
  40                 $this->mMainLanguageCode = $maincode;
  41                 $this->mVariants = $variants;
  42                 $this->mVariantFallbacks = $variantfallbacks;
  43                 $this->mCacheKey = $wgDBname . ":conversiontables";
  44                 $m = array('begin'=>'-{', 'flagsep'=>'|', 'codesep'=>':',
  45                                    'varsep'=>';', 'end'=>'}-');
  46                 $this->mMarkup = array_merge($m, $markup);
  47                 $f = array('A'=>'A', 'T'=>'T');
  48                 $this->mFlags = array_merge($f, $flags);
  49         }
  50
  51         /**
  52      * @access public
  53      */
  54         function getVariants() {
  55                 return $this->mVariants;
  56         }
  57
  58         /**
  59          * in case some variant is not defined in the markup, we need
  60          * to have some fallback. for example, in zh, normally people
  61          * will define zh-cn and zh-tw, but less so for zh-sg or zh-hk.
  62          * when zh-sg is preferred but not defined, we will pick zh-cn
  63          * in this case. right now this is only used by zh.
  64          *
  65          * @param string $v the language code of the variant
  66          * @return string the code of the fallback language or false if there is no fallback
  67      * @private
  68         */
  69         function getVariantFallback($v) {
  70                 return $this->mVariantFallbacks[$v];
  71         }
  72
  73
  74         /**
  75          * get preferred language variants.
  76          * @param boolean $fromUser Get it from $wgUser's preferences
  77      * @return string the preferred language code
  78      * @access public
  79         */
  80         function getPreferredVariant( $fromUser = true ) {
  81                 global $wgUser, $wgRequest;
  82
  83                 if($this->mPreferredVariant)
  84                         return $this->mPreferredVariant;
  85
  86                 // see if the preference is set in the request
  87                 $req = $wgRequest->getText( 'variant' );
  88                 if( in_array( $req, $this->mVariants ) ) {
  89                         $this->mPreferredVariant = $req;
  90                         return $req;
  91                 }
  92
  93                 // get language variant preference from logged in users
  94                 // Don't call this on stub objects because that causes infinite
  95                 // recursion during initialisation
  96                 if( $fromUser && $wgUser->isLoggedIn() )  {
  97                         $this->mPreferredVariant = $wgUser->getOption('variant');
  98                         return $this->mPreferredVariant;
  99                 }
 100
 101                 # FIXME rewrite code for parsing http header. The current code
 102                 # is written specific for detecting zh- variants
 103                 if( !$this->mPreferredVariant ) {
 104                         // see if some supported language variant is set in the
 105                         // http header, but we don't set the mPreferredVariant
 106                         // variable in case this is called before the user's
 107                         // preference is loaded
 108                         $pv=$this->mMainLanguageCode;
 109                         if(array_key_exists('HTTP_ACCEPT_LANGUAGE', $_SERVER)) {
 110                                 $header = str_replace( '_', '-', strtolower($_SERVER["HTTP_ACCEPT_LANGUAGE"]));
 111                                 $zh = strstr($header, 'zh-');
 112                                 if($zh) {
 113                                         $pv = substr($zh,0,5);
 114                                 }
 115                         }
 116                         return $pv;
 117                 }
 118         }
 119
 120         /**
 121      * dictionary-based conversion
 122      *
 123      * @param string $text the text to be converted
 124      * @param string $toVariant the target language code
 125      * @return string the converted text
 126      * @private
 127      */
 128         function autoConvert($text, $toVariant=false) {
 129                 $fname="LanguageConverter::autoConvert";
 130
 131                 wfProfileIn( $fname );
 132
 133                 if(!$this->mTablesLoaded)
 134                         $this->loadTables();
 135
 136                 if(!$toVariant)
 137                         $toVariant = $this->getPreferredVariant();
 138                 if(!in_array($toVariant, $this->mVariants))
 139                         return $text;
 140
 141                 /* we convert everything except:
 142                    1. html markups (anything between < and >)
 143                    2. html entities
 144                    3. place holders created by the parser
 145                 */
 146                 global $wgParser;
 147                 if (isset($wgParser))
 148                         $marker = '|' . $wgParser->UniqPrefix() . '[\-a-zA-Z0-9]+';
 149                 else
 150                         $marker = "";
 151
 152                 // this one is needed when the text is inside an html markup
 153                 $htmlfix = '|<[^>]+=\"[^(>=)]*$|^[^(<>=\")]*\"[^>]*>';
 154
 155                 $reg = '/<[^>]+>|&[a-z#][a-z0-9]+;' . $marker . $htmlfix . '/';
 156
 157                 $matches = preg_split($reg, $text, -1, PREG_SPLIT_OFFSET_CAPTURE);
 158
 159
 160                 $m = array_shift($matches);
 161                 $ret = strtr($m[0], $this->mTables[$toVariant]);
 162                 $mstart = $m[1]+strlen($m[0]);
 163                 foreach($matches as $m) {
 164                         $ret .= substr($text, $mstart, $m[1]-$mstart);
 165                         $ret .= strtr($m[0], $this->mTables[$toVariant]);
 166                         $mstart = $m[1] + strlen($m[0]);
 167                 }
 168                 wfProfileOut( $fname );
 169                 return $ret;
 170         }
 171
 172         /**
 173      * convert text to all supported variants
 174      *
 175      * @param string $text the text to be converted
 176      * @return array of string
 177      * @private
 178      */
 179         function autoConvertToAllVariants($text) {
 180                 $fname="LanguageConverter::autoConvertToAllVariants";
 181                 wfProfileIn( $fname );
 182                 if( !$this->mTablesLoaded )
 183                         $this->loadTables();
 184
 185                 $ret = array();
 186                 foreach($this->mVariants as $variant) {
 187                         $ret[$variant] = strtr($text, $this->mTables[$variant]);
 188                 }
 189                 wfProfileOut( $fname );
 190                 return $ret;
 191         }
 192
 193         /**
 194          * Convert text using a parser object for context
 195          */
 196         function parserConvert( $text, &$parser ) {
 197                 global $wgDisableLangConversion;
 198                 /* don't do anything if this is the conversion table */
 199                 if ( $parser->mTitle->getNamespace() == NS_MEDIAWIKI &&
 200                         strpos($parser->mTitle->getText, "Conversiontable") !== false )
 201                 {
 202                         return $text;
 203                 }
 204
 205                 if($wgDisableLangConversion)
 206                         return $text;
 207
 208                 $text = $this->convert( $text );
 209                 $parser->mOutput->setTitleText( $this->mTitleDisplay );
 210                 return $text;
 211         }
 212
 213         /**
 214          * convert text to different variants of a language. the automatic
 215          * conversion is done in autoConvert(). here we parse the text
 216          * marked with -{}-, which specifies special conversions of the
 217          * text that can not be accomplished in autoConvert()
 218          *
 219          * syntax of the markup:
 220          * -{code1:text1;code2:text2;...}-  or
 221          * -{text}- in which case no conversion should take place for text
 222      *
 223      * @param string $text text to be converted
 224      * @param bool $isTitle whether this conversion is for the article title
 225      * @return string converted text
 226      * @access public
 227      */
 228         function convert( $text , $isTitle=false) {
 229                 $mw =& MagicWord::get( 'notitleconvert'   );
 230                 if( $mw->matchAndRemove( $text ) )
 231                         $this->mDoTitleConvert = false;
 232
 233                 $mw =& MagicWord::get( 'nocontentconvert'   );
 234                 if( $mw->matchAndRemove( $text ) ) {
 235                         $this->mDoContentConvert = false;
 236                 }
 237
 238                 // no conversion if redirecting
 239                 $mw =& MagicWord::get( 'redirect'   );
 240                 if( $mw->matchStart( $text ))
 241                         return $text;
 242
 243                 if( $isTitle ) {
 244                         if( !$this->mDoTitleConvert ) {
 245                                 $this->mTitleDisplay = $text;
 246                                 return $text;
 247                         }
 248                         if( !empty($this->mTitleDisplay))
 249                                 return $this->mTitleDisplay;
 250
 251                         global $wgRequest;
 252                         $isredir = $wgRequest->getText( 'redirect', 'yes' );
 253                         $action = $wgRequest->getText( 'action' );
 254                         if ( $isredir == 'no' || $action == 'edit' ) {
 255                                 return $text;
 256                         }
 257                         else {
 258                                 $this->mTitleDisplay = $this->autoConvert($text);
 259                                 return $this->mTitleDisplay;
 260                         }
 261                 }
 262
 263                 if( !$this->mDoContentConvert )
 264                         return $text;
 265
 266                 $plang = $this->getPreferredVariant();
 267                 if( isset( $this->mVariantFallbacks[$plang] ) ) {
 268                         $fallback = $this->mVariantFallbacks[$plang];
 269                 } else {
 270                         // This sounds... bad?
 271                         $fallback = '';
 272                 }
 273
 274                 $tarray = explode($this->mMarkup['begin'], $text);
 275                 $tfirst = array_shift($tarray);
 276                 $text = $this->autoConvert($tfirst);
 277                 foreach($tarray as $txt) {
 278                         $marked = explode($this->mMarkup['end'], $txt, 2);
 279                         $flags = array();
 280                         $tt = explode($this->mMarkup['flagsep'], $marked[0], 2);
 281
 282                         if(sizeof($tt) == 2) {
 283                                 $f = explode($this->mMarkup['varsep'], $tt[0]);
 284                                 foreach($f as $ff) {
 285                                         $ff = trim($ff);
 286                                         if(array_key_exists($ff, $this->mFlags) &&
 287                                                 !array_key_exists($this->mFlags[$ff], $flags))
 288                                                 $flags[] = $this->mFlags[$ff];
 289                                 }
 290                                 $rules = $tt[1];
 291                         }
 292                         else
 293                                 $rules = $marked[0];
 294
 295 #FIXME: may cause trouble here...
 296                         //strip &nbsp; since it interferes with the parsing, plus,
 297                         //all spaces should be stripped in this tag anyway.
 298                         $rules = str_replace('&nbsp;', '', $rules);
 299
 300                         $carray = $this->parseManualRule($rules, $flags);
 301                         $disp = '';
 302                         if(array_key_exists($plang, $carray))
 303                                 $disp = $carray[$plang];
 304                         else if(array_key_exists($fallback, $carray))
 305                                 $disp = $carray[$fallback];
 306                         if($disp) {
 307                                 if(in_array('T',  $flags))
 308                                         $this->mTitleDisplay = $disp;
 309                                 else
 310                                         $text .= $disp;
 311
 312                                 if(in_array('A', $flags)) {
 313                                         /* modify the conversion table for this session*/
 314
 315                                         /* fill in the missing variants, if any,
 316                                             with fallbacks */
 317                                         foreach($this->mVariants as $v) {
 318                                                 if(!array_key_exists($v, $carray)) {
 319                                                         $vf = $this->getVariantFallback($v);
 320                                                         if(array_key_exists($vf, $carray))
 321                                                                 $carray[$v] = $carray[$vf];
 322                                                 }
 323                                         }
 324
 325                                         foreach($this->mVariants as $vfrom) {
 326                                                 if(!array_key_exists($vfrom, $carray))
 327                                                         continue;
 328                                                 foreach($this->mVariants as $vto) {
 329                                                         if($vfrom == $vto)
 330                                                                 continue;
 331                                                         if(!array_key_exists($vto, $carray))
 332                                                                 continue;
 333                                                         $this->mTables[$vto][$carray[$vfrom]] = $carray[$vto];
 334
 335                                                 }
 336                                         }
 337                                 }
 338                         }
 339                         else {
 340                                 $text .= $marked[0];
 341                         }
 342                         if(array_key_exists(1, $marked))
 343                                 $text .= $this->autoConvert($marked[1]);
 344                 }
 345
 346                 return $text;
 347         }
 348
 349         /**
 350          * parse the manually marked conversion rule
 351          * @param string $rule the text of the rule
 352          * @return array of the translation in each variant
 353          * @private
 354          */
 355         function parseManualRule($rules, $flags=array()) {
 356
 357                 $choice = explode($this->mMarkup['varsep'], $rules);
 358                 $carray = array();
 359                 if(sizeof($choice) == 1) {
 360                         /* a single choice */
 361                         foreach($this->mVariants as $v)
 362                                 $carray[$v] = $choice[0];
 363                 }
 364                 else {
 365                         foreach($choice as $c) {
 366                                 $v = explode($this->mMarkup['codesep'], $c);
 367                                 if(sizeof($v) != 2) // syntax error, skip
 368                                         continue;
 369                                 $carray[trim($v[0])] = trim($v[1]);
 370                         }
 371                 }
 372                 return $carray;
 373         }
 374
 375         /**
 376          * if a language supports multiple variants, it is
 377          * possible that non-existing link in one variant
 378          * actually exists in another variant. this function
 379          * tries to find it. See e.g. LanguageZh.php
 380          *
 381          * @param string $link the name of the link
 382          * @param mixed $nt the title object of the link
 383          * @return null the input parameters may be modified upon return
 384      * @access public
 385          */
 386         function findVariantLink( &$link, &$nt ) {
 387                 static $count=0; //used to limit this operation
 388                 static $cache=array();
 389                 global $wgDisableLangConversion;
 390                 $pref = $this->getPreferredVariant();
 391                 $ns=0;
 392                 if(is_object($nt))
 393                         $ns = $nt->getNamespace();
 394                 if( $count > 50 && $ns != NS_CATEGORY )
 395                         return;
 396                 $count++;
 397                 $variants = $this->autoConvertToAllVariants($link);
 398                 if($variants == false) //give up
 399                         return;
 400                 foreach( $variants as $v ) {
 401                         if(isset($cache[$v]))
 402                                 continue;
 403                         $cache[$v] = 1;
 404                         $varnt = Title::newFromText( $v, $ns );
 405                         if( $varnt && $varnt->getArticleID() > 0 ) {
 406                                 $nt = $varnt;
 407                                 if( !$wgDisableLangConversion )
 408                                         $link = $v;
 409                                 break;
 410                         }
 411                 }
 412         }
 413
 414     /**
 415      * returns language specific hash options
 416      *
 417      * @access public
 418      */
 419         function getExtraHashOptions() {
 420                 $variant = $this->getPreferredVariant();
 421                 return '!' . $variant ;
 422         }
 423
 424     /**
 425      * get title text as defined in the body of the article text
 426      *
 427      * @access public
 428      */
 429         function getParsedTitle() {
 430                 return $this->mTitleDisplay;
 431         }
 432
 433         /**
 434      * a write lock to the cache
 435      *
 436      * @private
 437      */
 438         function lockCache() {
 439                 global $wgMemc;
 440                 $success = false;
 441                 for($i=0; $i<30; $i++) {
 442                         if($success = $wgMemc->add($this->mCacheKey . "lock", 1, 10))
 443                                 break;
 444                         sleep(1);
 445                 }
 446                 return $success;
 447         }
 448
 449         /**
 450      * unlock cache
 451      *
 452      * @private
 453      */
 454         function unlockCache() {
 455                 global $wgMemc;
 456                 $wgMemc->delete($this->mCacheKey . "lock");
 457         }
 458
 459
 460         /**
 461      * Load default conversion tables
 462      * This method must be implemented in derived class
 463      *
 464      * @private
 465      */
 466         function loadDefaultTables() {
 467                 $name = get_class($this);
 468                 wfDie("Must implement loadDefaultTables() method in class $name");
 469         }
 470
 471         /**
 472      * load conversion tables either from the cache or the disk
 473      * @private
 474      */
 475         function loadTables($fromcache=true) {
 476                 global $wgMemc;
 477                 if( $this->mTablesLoaded )
 478                         return;
 479                 $this->mTablesLoaded = true;
 480                 if($fromcache) {
 481                         $this->mTables = $wgMemc->get( $this->mCacheKey );
 482                         if( !empty( $this->mTables ) ) //all done
 483                                 return;
 484                 }
 485                 // not in cache, or we need a fresh reload.
 486                 // we will first load the default tables
 487                 // then update them using things in MediaWiki:Zhconversiontable/*
 488                 global $wgMessageCache;
 489                 $this->loadDefaultTables();
 490                 foreach($this->mVariants as $var) {
 491                         $cached = $this->parseCachedTable($var);
 492                         $this->mTables[$var] = array_merge($this->mTables[$var], $cached);
 493                 }
 494
 495                 $this->postLoadTables();
 496
 497                 if($this->lockCache()) {
 498                         $wgMemc->set($this->mCacheKey, $this->mTables, 43200);
 499                         $this->unlockCache();
 500                 }
 501         }
 502
 503     /**
 504      * Hook for post processig after conversion tables are loaded
 505      *
 506      */
 507         function postLoadTables() {}
 508
 509     /**
 510      * Reload the conversion tables
 511      *
 512      * @private
 513      */
 514         function reloadTables() {
 515                 if($this->mTables)
 516                         unset($this->mTables);
 517                 $this->mTablesLoaded = false;
 518                 $this->loadTables(false);
 519         }
 520
 521
 522         /**
 523      * parse the conversion table stored in the cache
 524      *
 525      * the tables should be in blocks of the following form:
 526
 527      *          -{
 528      *                  word => word ;
 529      *                  word => word ;
 530      *                  ...
 531      *          }-
 532      *
 533      *  to make the tables more manageable, subpages are allowed
 534      *  and will be parsed recursively if $recursive=true
 535      *
 536      * @private
 537          */
 538         function parseCachedTable($code, $subpage='', $recursive=true) {
 539                 global $wgMessageCache;
 540                 static $parsed = array();
 541
 542                 if(!is_object($wgMessageCache))
 543                         return array();
 544
 545                 $key = 'Conversiontable/'.$code;
 546                 if($subpage)
 547                         $key .= '/' . $subpage;
 548
 549                 if(array_key_exists($key, $parsed))
 550                         return array();
 551
 552
 553                 $txt = $wgMessageCache->get( $key, true, true, true );
 554
 555                 // get all subpage links of the form
 556                 // [[MediaWiki:conversiontable/zh-xx/...|...]]
 557                 $linkhead = $this->mLangObj->getNsText(NS_MEDIAWIKI) . ':Conversiontable';
 558                 $subs = explode('[[', $txt);
 559                 $sublinks = array();
 560                 foreach( $subs as $sub ) {
 561                         $link = explode(']]', $sub, 2);
 562                         if(count($link) != 2)
 563                                 continue;
 564                         $b = explode('|', $link[0]);
 565                         $b = explode('/', trim($b[0]), 3);
 566                         if(count($b)==3)
 567                                 $sublink = $b[2];
 568                         else
 569                                 $sublink = '';
 570
 571                         if($b[0] == $linkhead && $b[1] == $code) {
 572                                 $sublinks[] = $sublink;
 573                         }
 574                 }
 575
 576
 577                 // parse the mappings in this page
 578                 $blocks = explode($this->mMarkup['begin'], $txt);
 579                 array_shift($blocks);
 580                 $ret = array();
 581                 foreach($blocks as $block) {
 582                         $mappings = explode($this->mMarkup['end'], $block, 2);
 583                         $stripped = str_replace(array("'", '"', '*','#'), '', $mappings[0]);
 584                         $table = explode( ';', $stripped );
 585                         foreach( $table as $t ) {
 586                                 $m = explode( '=>', $t );
 587                                 if( count( $m ) != 2)
 588                                         continue;
 589                                 // trim any trailling comments starting with '//'
 590                                 $tt = explode('//', $m[1], 2);
 591                                 $ret[trim($m[0])] = trim($tt[0]);
 592                         }
 593                 }
 594                 $parsed[$key] = true;
 595
 596
 597                 // recursively parse the subpages
 598                 if($recursive) {
 599                         foreach($sublinks as $link) {
 600                                 $s = $this->parseCachedTable($code, $link, $recursive);
 601                                 $ret = array_merge($ret, $s);
 602                         }
 603                 }
 604
 605                 if ($this->mUcfirst) {
 606                         foreach ($ret as $k => $v) {
 607                                 $ret[LanguageUtf8::ucfirst($k)] = LanguageUtf8::ucfirst($v);
 608                         }
 609                 }
 610                 return $ret;
 611         }
 612
 613         /**
 614          * Enclose a string with the "no conversion" tag. This is used by
 615          * various functions in the Parser
 616          *
 617          * @param string $text text to be tagged for no conversion
 618          * @return string the tagged text
 619         */
 620         function markNoConversion($text) {
 621                 # don't mark if already marked
 622                 if(strpos($text, $this->mMarkup['begin']) ||
 623                    strpos($text, $this->mMarkup['end']))
 624                         return $text;
 625
 626                 $ret = $this->mMarkup['begin'] . $text . $this->mMarkup['end'];
 627                 return $ret;
 628         }
 629
 630         /**
 631          * convert the sorting key for category links. this should make different
 632          * keys that are variants of each other map to the same key
 633         */
 634         function convertCategoryKey( $key ) {
 635                 return $key;
 636         }
 637         /**
 638      * hook to refresh the cache of conversion tables when
 639      * MediaWiki:conversiontable* is updated
 640      * @private
 641         */
 642         function OnArticleSaveComplete($article, $user, $text, $summary, $isminor, $iswatch, $section) {
 643                 $titleobj = $article->getTitle();
 644                 if($titleobj->getNamespace() == NS_MEDIAWIKI) {
 645             /*
 646                         global $wgContLang; // should be an LanguageZh.
 647                         if(get_class($wgContLang) != 'languagezh')
 648                                 return true;
 649             */
 650                         $title = $titleobj->getDBkey();
 651                         $t = explode('/', $title, 3);
 652                         $c = count($t);
 653                         if( $c > 1 && $t[0] == 'Conversiontable' ) {
 654                                 if(in_array($t[1], $this->mVariants)) {
 655                                         $this->reloadTables();
 656                                 }
 657                         }
 658                 }
 659                 return true;
 660         }
 661 }
 662
 663 ?>