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