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 $flags array defining the custom strings that maps to the flags
  31      * @access public
  32      */
  33         function LanguageConverter($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 $v the language code of the variant
  66          * @return string the code of the fallback language or false if there is no fallback
  67      * @access private
  68         */
  69         function getVariantFallback($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 $toVariant the target language code
 122      * @return string the converted text
 123      * @access private
 124      */
 125         function autoConvert($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                 $reg = '/<[^>]+>|&[a-z#][a-z0-9]+;' . $marker . '/';
 149                 $matches = preg_split($reg, $text, -1, PREG_SPLIT_OFFSET_CAPTURE);
 150
 151
 152                 $m = array_shift($matches);
 153                 $ret = strtr($m[0], $this->mTables[$toVariant]);
 154                 $mstart = $m[1]+strlen($m[0]);
 155                 foreach($matches as $m) {
 156                         $ret .= substr($text, $mstart, $m[1]-$mstart);
 157                         $ret .= strtr($m[0], $this->mTables[$toVariant]);
 158                         $mstart = $m[1] + strlen($m[0]);
 159                 }
 160                 wfProfileOut( $fname );
 161                 return $ret;
 162         }
 163
 164         /**
 165      * convert text to all supported variants
 166      *
 167      * @param string $text the text to be converted
 168      * @return array of string
 169      * @access private
 170      */
 171         function autoConvertToAllVariants($text) {
 172                 $fname="LanguageConverter::autoConvertToAllVariants";
 173                 wfProfileIn( $fname );
 174                 if( !$this->mTablesLoaded )
 175                         $this->loadTables();
 176
 177                 $ret = array();
 178                 foreach($this->mVariants as $variant) {
 179                         $ret[$variant] = strtr($text, $this->mTables[$variant]);
 180                 }
 181                 wfProfileOut( $fname );
 182                 return $ret;
 183         }
 184
 185         /**
 186          * convert text to different variants of a language. the automatic
 187          * conversion is done in autoConvert(). here we parse the text
 188          * marked with -{}-, which specifies special conversions of the
 189          * text that can not be accomplished in autoConvert()
 190          *
 191          * syntax of the markup:
 192          * -{code1:text1;code2:text2;...}-  or
 193          * -{text}- in which case no conversion should take place for text
 194      *
 195      * @param string $text text to be converted
 196      * @param bool $isTitle whether this conversion is for the article title
 197      * @return string converted text
 198      * @access public
 199      */
 200         function convert( $text , $isTitle=false) {
 201                 global $wgDisableLangConversion;
 202                 global $wgTitle;
 203
 204                 /* don't do anything if this is the conversion table */
 205                 if($wgTitle->getNamespace() == NS_MEDIAWIKI &&
 206                    strpos($wgTitle->getText(), "Conversiontable")!==false)
 207                         return $text;
 208
 209                 if($wgDisableLangConversion)
 210                         return $text;
 211
 212                 $mw =& MagicWord::get( MAG_NOTITLECONVERT );
 213                 if( $mw->matchAndRemove( $text ) )
 214                         $this->mDoTitleConvert = false;
 215
 216                 $mw =& MagicWord::get( MAG_NOCONTENTCONVERT );
 217                 if( $mw->matchAndRemove( $text ) ) {
 218                         $this->mDoContentConvert = false;
 219                 }
 220
 221                 // no conversion if redirecting
 222                 $mw =& MagicWord::get( MAG_REDIRECT );
 223                 if( $mw->matchStart( $text ))
 224                         return $text;
 225
 226                 if( $isTitle ) {
 227                         if( !$this->mDoTitleConvert ) {
 228                                 $this->mTitleDisplay = $text;
 229                                 return $text;
 230                         }
 231                         if( !empty($this->mTitleDisplay))
 232                                 return $this->mTitleDisplay;
 233
 234                         global $wgRequest;
 235                         $isredir = $wgRequest->getText( 'redirect', 'yes' );
 236                         $action = $wgRequest->getText( 'action' );
 237                         if ( $isredir == 'no' || $action == 'edit' ) {
 238                                 return $text;
 239                         }
 240                         else {
 241                                 $this->mTitleDisplay = $this->autoConvert($text);
 242                                 return $this->mTitleDisplay;
 243                         }
 244                 }
 245
 246                 if( !$this->mDoContentConvert )
 247                         return $text;
 248
 249                 $plang = $this->getPreferredVariant();
 250                 $fallback = $this->mVariantFallbacks[$plang];
 251
 252                 $tarray = explode($this->mMarkup['begin'], $text);
 253                 $tfirst = array_shift($tarray);
 254                 $text = $this->autoConvert($tfirst);
 255                 foreach($tarray as $txt) {
 256                         $marked = explode($this->mMarkup['end'], $txt);
 257                         $flags = array();
 258                         $tt = explode($this->mMarkup['flagsep'], $marked[0], 2);
 259
 260                         if(sizeof($tt) == 2) {
 261                                 $f = explode($this->mMarkup['varsep'], $tt[0]);
 262                                 foreach($f as $ff) {
 263                                         $ff = trim($ff);
 264                                         if(array_key_exists($ff, $this->mFlags) &&
 265                                                 !array_key_exists($this->mFlags[$ff], $flags))
 266                                                 $flags[] = $this->mFlags[$ff];
 267                                 }
 268                                 $rules = $tt[1];
 269                         }
 270                         else
 271                                 $rules = $marked[0];
 272
 273 #FIXME: may cause trouble here...
 274                         //strip &nbsp; since it interferes with the parsing, plus,
 275                         //all spaces should be stripped in this tag anyway.
 276                         $rules = str_replace('&nbsp;', '', $rules);
 277
 278                         $carray = $this->parseManualRule($rules, $flags);
 279                         $disp = '';
 280                         if(array_key_exists($plang, $carray))
 281                                 $disp = $carray[$plang];
 282                         else if(array_key_exists($fallback, $carray))
 283                                 $disp = $carray[$fallback];
 284                         if($disp) {
 285                                 if(in_array('T',  $flags))
 286                                         $this->mTitleDisplay = $disp;
 287                                 else
 288                                         $text .= $disp;
 289
 290                                 if(in_array('A', $flags)) {
 291                                         /* modify the conversion table for this session*/
 292
 293                                         /* fill in the missing variants, if any,
 294                                             with fallbacks */
 295                                         foreach($this->mVariants as $v) {
 296                                                 if(!array_key_exists($v, $carray)) {
 297                                                         $vf = $this->getVariantFallback($v);
 298                                                         if(array_key_exists($vf, $carray))
 299                                                                 $carray[$v] = $carray[$vf];
 300                                                 }
 301                                         }
 302
 303                                         foreach($this->mVariants as $vfrom) {
 304                                                 if(!array_key_exists($vfrom, $carray))
 305                                                         continue;
 306                                                 foreach($this->mVariants as $vto) {
 307                                                         if($vfrom == $vto)
 308                                                                 continue;
 309                                                         if(!array_key_exists($vto, $carray))
 310                                                                 continue;
 311                                                         $this->mTables[$vto][$carray[$vfrom]] = $carray[$vto];
 312
 313                                                 }
 314                                         }
 315                                 }
 316                         }
 317                         else {
 318                                 $text .= $marked[0];
 319                         }
 320                         if(array_key_exists(1, $marked))
 321                                 $text .= $this->autoConvert($marked[1]);
 322                 }
 323
 324                 return $text;
 325         }
 326
 327         /**
 328          * parse the manually marked conversion rule
 329          * @param string $rule the text of the rule
 330          * @return array of the translation in each variant
 331          * @access private
 332          */
 333         function parseManualRule($rules, $flags=array()) {
 334
 335                 $choice = explode($this->mMarkup['varsep'], $rules);
 336                 $carray = array();
 337                 if(sizeof($choice) == 1) {
 338                         /* a single choice */
 339                         foreach($this->mVariants as $v)
 340                                 $carray[$v] = $choice[0];
 341                 }
 342                 else {
 343                         foreach($choice as $c) {
 344                                 $v = explode($this->mMarkup['codesep'], $c);
 345                                 if(sizeof($v) != 2) // syntax error, skip
 346                                         continue;
 347                                 $carray[trim($v[0])] = trim($v[1]);
 348                         }
 349                 }
 350                 return $carray;
 351         }
 352
 353         /**
 354          * if a language supports multiple variants, it is
 355          * possible that non-existing link in one variant
 356          * actually exists in another variant. this function
 357          * tries to find it. See e.g. LanguageZh.php
 358          *
 359          * @param string $link the name of the link
 360          * @param mixed $nt the title object of the link
 361          * @return null the input parameters may be modified upon return
 362      * @access public
 363          */
 364         function findVariantLink( &$link, &$nt ) {
 365                 static $count=0; //used to limit this operation
 366                 static $cache=array();
 367                 global $wgDisableLangConversion;
 368                 $pref = $this->getPreferredVariant();
 369                 $ns=0;
 370                 if(is_object($nt))
 371                         $ns = $nt->getNamespace();
 372                 if( $count > 50 && $ns != NS_CATEGORY )
 373                         return;
 374                 $count++;
 375                 $variants = $this->autoConvertToAllVariants($link);
 376                 if($variants == false) //give up
 377                         return;
 378                 foreach( $variants as $v ) {
 379                         if(isset($cache[$v]))
 380                                 continue;
 381                         $cache[$v] = 1;
 382                         $varnt = Title::newFromText( $v );
 383                         if( $varnt && $varnt->getArticleID() > 0 ) {
 384                                 $nt = $varnt;
 385                                 if( !$wgDisableLangConversion )
 386                                         $link = $v;
 387                                 break;
 388                         }
 389                 }
 390         }
 391
 392     /**
 393      * returns language specific hash options
 394      *
 395      * @access public
 396      */
 397         function getExtraHashOptions() {
 398                 $variant = $this->getPreferredVariant();
 399                 return '!' . $variant ;
 400         }
 401
 402     /**
 403      * get title text as defined in the body of the article text
 404      *
 405      * @access public
 406      */
 407         function getParsedTitle() {
 408                 return $this->mTitleDisplay;
 409         }
 410
 411         /**
 412      * a write lock to the cache
 413      *
 414      * @access private
 415      */
 416         function lockCache() {
 417                 global $wgMemc;
 418                 $success = false;
 419                 for($i=0; $i<30; $i++) {
 420                         if($success = $wgMemc->add($this->mCacheKey . "lock", 1, 10))
 421                                 break;
 422                         sleep(1);
 423                 }
 424                 return $success;
 425         }
 426
 427         /**
 428      * unlock cache
 429      *
 430      * @access private
 431      */
 432         function unlockCache() {
 433                 global $wgMemc;
 434                 $wgMemc->delete($this->mCacheKey . "lock");
 435         }
 436
 437
 438         /**
 439      * Load default conversion tables
 440      * This method must be implemented in derived class
 441      *
 442      * @access private
 443      */
 444         function loadDefaultTables() {
 445                 $name = get_class($this);
 446                 die("Must implement loadDefaultTables() method in class $name");
 447         }
 448
 449         /**
 450      * load conversion tables either from the cache or the disk
 451      * @access private
 452      */
 453         function loadTables($fromcache=true) {
 454                 global $wgMemc;
 455                 if( $this->mTablesLoaded )
 456                         return;
 457                 $this->mTablesLoaded = true;
 458                 if($fromcache) {
 459                         $this->mTables = $wgMemc->get( $this->mCacheKey );
 460                         if( !empty( $this->mTables ) ) //all done
 461                                 return;
 462                 }
 463                 // not in cache, or we need a fresh reload.
 464                 // we will first load the default tables
 465                 // then update them using things in MediaWiki:Zhconversiontable/*
 466                 global $wgMessageCache;
 467                 $this->loadDefaultTables();
 468                 foreach($this->mVariants as $var) {
 469                         $cached = $this->parseCachedTable($var);
 470                         $this->mTables[$var] = array_merge($this->mTables[$var], $cached);
 471                 }
 472
 473                 $this->postLoadTables();
 474
 475                 if($this->lockCache()) {
 476                         $wgMemc->set($this->mCacheKey, $this->mTables, 43200);
 477                         $this->unlockCache();
 478                 }
 479         }
 480
 481     /**
 482      * Hook for post processig after conversion tables are loaded
 483      *
 484      */
 485         function postLoadTables() {}
 486
 487     /**
 488      * Reload the conversion tables
 489      *
 490      * @access private
 491      */
 492         function reloadTables() {
 493                 if($this->mTables)
 494                         unset($this->mTables);
 495                 $this->mTablesLoaded = false;
 496                 $this->loadTables(false);
 497         }
 498
 499
 500         /**
 501      * parse the conversion table stored in the cache
 502      *
 503      * the tables should be in blocks of the following form:
 504
 505      *          -{
 506      *                  word => word ;
 507      *                  word => word ;
 508      *                  ...
 509      *          }-
 510      *
 511      *  to make the tables more manageable, subpages are allowed
 512      *  and will be parsed recursively if $recursive=true
 513      *
 514      * @access private
 515          */
 516         function parseCachedTable($code, $subpage='', $recursive=true) {
 517                 global $wgMessageCache;
 518                 static $parsed = array();
 519
 520                 if(!is_object($wgMessageCache))
 521                         return array();
 522
 523                 $key = 'Conversiontable/'.$code;
 524                 if($subpage)
 525                         $key .= '/' . $subpage;
 526
 527                 if(array_key_exists($key, $parsed))
 528                         return array();
 529
 530
 531                 $txt = $wgMessageCache->get( $key, true, true, true );
 532
 533                 // get all subpage links of the form
 534                 // [[MediaWiki:conversiontable/zh-xx/...|...]]
 535                 $linkhead = $this->mLangObj->getNsText(NS_MEDIAWIKI) . ':Conversiontable';
 536                 $subs = explode('[[', $txt);
 537                 $sublinks = array();
 538                 foreach( $subs as $sub ) {
 539                         $link = explode(']]', $sub, 2);
 540                         if(count($link) != 2)
 541                                 continue;
 542                         $b = explode('|', $link[0]);
 543                         $b = explode('/', trim($b[0]), 3);
 544                         if(count($b)==3)
 545                                 $sublink = $b[2];
 546                         else
 547                                 $sublink = '';
 548
 549                         if($b[0] == $linkhead && $b[1] == $code) {
 550                                 $sublinks[] = $sublink;
 551                         }
 552                 }
 553
 554
 555                 // parse the mappings in this page
 556                 $blocks = explode($this->mMarkup['begin'], $txt);
 557                 array_shift($blocks);
 558                 $ret = array();
 559                 foreach($blocks as $block) {
 560                         $mappings = explode($this->mMarkup['end'], $block, 2);
 561                         $stripped = str_replace(array("'", '"', '*','#'), '', $mappings[0]);
 562                         $table = explode( ';', $stripped );
 563                         foreach( $table as $t ) {
 564                                 $m = explode( '=>', $t );
 565                                 if( count( $m ) != 2)
 566                                         continue;
 567                                 // trim any trailling comments starting with '//'
 568                                 $tt = explode('//', $m[1], 2);
 569                                 $ret[trim($m[0])] = trim($tt[0]);
 570                         }
 571                 }
 572                 $parsed[$key] = true;
 573
 574
 575                 // recursively parse the subpages
 576                 if($recursive) {
 577                         foreach($sublinks as $link) {
 578                                 $s = $this->parseCachedTable($code, $link, $recursive);
 579                                 $ret = array_merge($ret, $s);
 580                         }
 581                 }
 582
 583                 if ($this->mUcfirst) {
 584                         foreach ($ret as $k => $v) {
 585                                 $ret[LanguageUtf8::ucfirst($k)] = LanguageUtf8::ucfirst($v);
 586                         }
 587                 }
 588                 return $ret;
 589         }
 590
 591         /**
 592          * Enclose a string with the "no conversion" tag. This is used by
 593          * various functions in the Parser
 594          *
 595          * @param string $text text to be tagged for no conversion
 596          * @return string the tagged text
 597         */
 598         function markNoConversion($text) {
 599                 # don't mark if already marked
 600                 if(strpos($text, $this->mMarkup['begin']) ||
 601                    strpos($text, $this->mMarkup['end']))
 602                         return $text;
 603
 604                 $ret = $this->mMarkup['begin'] . $text . $this->mMarkup['end'];
 605                 return $ret;
 606         }
 607
 608         /**
 609          * convert the sorting key for category links. this should make different
 610          * keys that are variants of each other map to the same key
 611         */
 612         function convertCategoryKey( $key ) {
 613                 return $key;
 614         }
 615         /**
 616      * hook to refresh the cache of conversion tables when
 617      * MediaWiki:conversiontable* is updated
 618      * @access private
 619         */
 620         function OnArticleSaveComplete($article, $user, $text, $summary, $isminor, $iswatch, $section) {
 621                 $titleobj = $article->getTitle();
 622                 if($titleobj->getNamespace() == NS_MEDIAWIKI) {
 623             /*
 624                         global $wgContLang; // should be an LanguageZh.
 625                         if(get_class($wgContLang) != 'languagezh')
 626                                 return true;
 627             */
 628                         $title = $titleobj->getDBkey();
 629                         $t = explode('/', $title, 3);
 630                         $c = count($t);
 631                         if( $c > 1 && $t[0] == 'Conversiontable' ) {
 632                                 if(in_array($t[1], $this->mVariants)) {
 633                                         $this->reloadTables();
 634                                 }
 635                         }
 636                 }
 637                 return true;
 638         }
 639 }
 640
 641 ?>