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