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