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