languages/LanguageConverter.php

   1 <?php
   2
   3 /**
   4  * Contains the LanguageConverter class and ConverterRule class
   5  * @ingroup Language
   6  *
   7  * @license http://www.gnu.org/copyleft/gpl.html GNU General Public License
   8  * @file
   9  */
  10
  11 /**
  12  * base class for language convert
  13  * @ingroup Language
  14  *
  15  * @author Zhengzhu Feng <zhengzhu@gmail.com>
  16  * @maintainers fdcn <fdcn64@gmail.com>, shinjiman <shinjiman@gmail.com>, PhiLiP <philip.npc@gmail.com>
  17  */
  18 class LanguageConverter {
  19         var $mPreferredVariant='';
  20         var $mMainLanguageCode;
  21         var $mVariants, $mVariantFallbacks, $mVariantNames;
  22         var $mTablesLoaded = false;
  23         var $mTables;
  24         var $mManualAddTables;
  25         var $mManualRemoveTables;
  26         var $mNamespaceTables;
  27         var $mTitleDisplay='';
  28         var $mDoTitleConvert=true, $mDoContentConvert=true;
  29         var $mManualLevel; // 'bidirectional' 'unidirectional' 'disable' for each variants
  30         var $mTitleFromFlag = false;
  31         var $mCacheKey;
  32         var $mLangObj;
  33         var $mMarkup;
  34         var $mFlags;
  35         var $mDescCodeSep = ':',$mDescVarSep = ';';
  36         var $mUcfirst = false;
  37
  38         const CACHE_VERSION_KEY = 'VERSION 6';
  39
  40         /**
  41          * Constructor
  42          *
  43          * @param string $maincode the main language code of this language
  44          * @param array $variants the supported variants of this language
  45          * @param array $variantfallback the fallback language of each variant
  46          * @param array $markup array defining the markup used for manual conversion
  47          * @param array $flags array defining the custom strings that maps to the flags
  48          * @param array $manualLevel limit for supported variants
  49          * @public
  50          */
  51         function __construct($langobj, $maincode,
  52                                                                 $variants=array(),
  53                                                                 $variantfallbacks=array(),
  54                                                                 $markup=array(),
  55                                                                 $flags = array(),
  56                                                                 $manualLevel = array() ) {
  57                 $this->mLangObj = $langobj;
  58                 $this->mMainLanguageCode = $maincode;
  59                 $this->mVariants = $variants;
  60                 $this->mVariantFallbacks = $variantfallbacks;
  61                 global $wgLanguageNames;
  62                 $this->mVariantNames = $wgLanguageNames;
  63                 $this->mCacheKey = wfMemcKey( 'conversiontables', $maincode );
  64                 $m = array(
  65                         'begin'=>'-{',
  66                         'flagsep'=>'|',
  67                         'unidsep'=>'=>', //for unidirectional conversion
  68                         'codesep'=>':',
  69                         'varsep'=>';',
  70                         'end'=>'}-'
  71                 );
  72                 $this->mMarkup = array_merge($m, $markup);
  73                 $f = array(
  74                         // 'S' show converted text
  75                         // '+' add rules for alltext
  76                         // 'E' the gave flags is error
  77                         // these flags above are reserved for program
  78                         'A'=>'A',       // add rule for convert code (all text convert)
  79                         'T'=>'T',       // title convert
  80                         'R'=>'R',       // raw content
  81                         'D'=>'D',       // convert description (subclass implement)
  82                         '-'=>'-',       // remove convert (not implement)
  83                         'H'=>'H',       // add rule for convert code (but no display in placed code )
  84                         'N'=>'N'        // current variant name
  85                 );
  86                 $this->mFlags = array_merge($f, $flags);
  87                 foreach( $this->mVariants as $v) {
  88                         $this->mManualLevel[$v]=array_key_exists($v,$manualLevel)
  89                                                                 ?$manualLevel[$v]
  90                                                                 :'bidirectional';
  91                         $this->mManualAddTables[$v] = array();
  92                         $this->mManualRemoveTables[$v] = array();
  93                         $this->mNamespaceTables[$v] = array();
  94                         $this->mFlags[$v] = $v;
  95                 }
  96         }
  97
  98         /**
  99          * @public
 100          */
 101         function getVariants() {
 102                 return $this->mVariants;
 103         }
 104
 105         /**
 106          * in case some variant is not defined in the markup, we need
 107          * to have some fallback. for example, in zh, normally people
 108          * will define zh-hans and zh-hant, but less so for zh-sg or zh-hk.
 109          * when zh-sg is preferred but not defined, we will pick zh-hans
 110          * in this case. right now this is only used by zh.
 111          *
 112          * @param string $v the language code of the variant
 113          * @return string array the code of the fallback language or false if there is no fallback
 114          * @public
 115          */
 116         function getVariantFallbacks($v) {
 117                 if( isset( $this->mVariantFallbacks[$v] ) ) {
 118                         return $this->mVariantFallbacks[$v];
 119                 }
 120                 return $this->mMainLanguageCode;
 121         }
 122
 123         /**
 124          * get preferred language variants.
 125          * @param boolean $fromUser Get it from $wgUser's preferences
 126          * @return string the preferred language code
 127          * @public
 128          */
 129         function getPreferredVariant( $fromUser = true ) {
 130                 global $wgUser, $wgRequest, $wgVariantArticlePath, $wgDefaultLanguageVariant;
 131
 132                 if($this->mPreferredVariant)
 133                         return $this->mPreferredVariant;
 134
 135                 // figure out user lang without constructing wgLang to avoid infinite recursion
 136                 if( $fromUser )
 137                         $defaultUserLang = $wgUser->getOption( 'language' );
 138                 else
 139                         $defaultUserLang = $this->mMainLanguageCode;
 140                 $userLang = $wgRequest->getVal( 'uselang', $defaultUserLang );
 141                 // see if interface language is same as content, if not, prevent conversion
 142                 if( ! in_array( $userLang, $this->mVariants ) ){
 143                         $this->mPreferredVariant = $this->mMainLanguageCode; // no conversion
 144                         return $this->mPreferredVariant;
 145                 }
 146
 147                 // see if the preference is set in the request
 148                 $req = $wgRequest->getText( 'variant' );
 149                 if( in_array( $req, $this->mVariants ) ) {
 150                         $this->mPreferredVariant = $req;
 151                         return $req;
 152                 }
 153
 154                 // check the syntax /code/ArticleTitle
 155                 if($wgVariantArticlePath!=false && isset($_SERVER['SCRIPT_NAME'])){
 156                         // Note: SCRIPT_NAME probably won't hold the correct value if PHP is run as CGI
 157                         // (it will hold path to php.cgi binary), and might not exist on some very old PHP installations
 158                         $scriptBase = basename( $_SERVER['SCRIPT_NAME'] );
 159                         if(in_array($scriptBase,$this->mVariants)){
 160                                 $this->mPreferredVariant = $scriptBase;
 161                                 return $this->mPreferredVariant;
 162                         }
 163                 }
 164
 165                 // get language variant preference from logged in users
 166                 // Don't call this on stub objects because that causes infinite
 167                 // recursion during initialisation
 168                 if( $fromUser && $wgUser->isLoggedIn() )  {
 169                         $this->mPreferredVariant = $wgUser->getOption('variant');
 170                         return $this->mPreferredVariant;
 171                 }
 172
 173                 // see if default variant is globaly set
 174                 if($wgDefaultLanguageVariant != false  &&  in_array( $wgDefaultLanguageVariant, $this->mVariants )){
 175                         $this->mPreferredVariant = $wgDefaultLanguageVariant;
 176                         return $this->mPreferredVariant;
 177                 }
 178
 179                 # FIXME rewrite code for parsing http header. The current code
 180                 # is written specific for detecting zh- variants
 181                 if( !$this->mPreferredVariant ) {
 182                         // see if some supported language variant is set in the
 183                         // http header, but we don't set the mPreferredVariant
 184                         // variable in case this is called before the user's
 185                         // preference is loaded
 186                         $pv=$this->mMainLanguageCode;
 187                         if(array_key_exists('HTTP_ACCEPT_LANGUAGE', $_SERVER)) {
 188                                 $header = str_replace( '_', '-', strtolower($_SERVER["HTTP_ACCEPT_LANGUAGE"]));
 189                                 $zh = strstr($header, $pv.'-');
 190                                 if($zh) {
 191                                         $ary = split("[,;]",$zh);
 192                                         $pv = $ary[0];
 193                                 }
 194                         }
 195                         // don't try to return bad variant
 196                         if(in_array( $pv, $this->mVariants ))
 197                                 return $pv;
 198                 }
 199
 200                 return $this->mMainLanguageCode;
 201
 202         }
 203
 204         /**
 205          * caption convert, base on preg_replace_callback
 206          *
 207          * to convert text in "title" or "alt", like '<img alt="text" ... '
 208          * or '<span title="text" ... '
 209          *
 210          * @return string like ' alt="yyyy"' or ' title="yyyy"'
 211          * @private
 212          */
 213         function captionConvert( $matches ) {
 214                 $toVariant = $this->getPreferredVariant();
 215                 $title = $matches[1];
 216                 $text  = $matches[2];
 217                 // we convert captions except URL
 218                 if( !strpos( $text, '://' ) )
 219                         $text = $this->translate($text, $toVariant);
 220                 return " $title=\"$text\"";
 221         }
 222
 223         /**
 224          * dictionary-based conversion
 225          *
 226          * @param string $text the text to be converted
 227          * @param string $toVariant the target language code
 228          * @return string the converted text
 229          * @private
 230          */
 231         function autoConvert($text, $toVariant=false) {
 232                 $fname="LanguageConverter::autoConvert";
 233
 234                 wfProfileIn( $fname );
 235
 236                 if(!$this->mTablesLoaded)
 237                         $this->loadTables();
 238
 239                 if(!$toVariant)
 240                         $toVariant = $this->getPreferredVariant();
 241                 if(!in_array($toVariant, $this->mVariants))
 242                         return $text;
 243
 244                 /* we convert everything except:
 245                    1. html markups (anything between < and >)
 246                    2. html entities
 247                    3. place holders created by the parser
 248                 */
 249                 global $wgParser;
 250                 if (isset($wgParser) && $wgParser->UniqPrefix()!=''){
 251                         $marker = '|' . $wgParser->UniqPrefix() . '[\-a-zA-Z0-9]+';
 252                 } else
 253                         $marker = "";
 254
 255                 // this one is needed when the text is inside an html markup
 256                 $htmlfix = '|<[^>]+$|^[^<>]*>';
 257
 258                 // disable convert to variants between <code></code> tags
 259                 $codefix = '<code>.+?<\/code>|';
 260                 // disable convertsion of <script type="text/javascript"> ... </script>
 261                 $scriptfix = '<script.*?>.*?<\/script>|';
 262                 // disable conversion of <pre xxxx> ... </pre>
 263                 $prefix = '<pre.*?>.*?<\/pre>|';
 264
 265                 $reg = '/'.$codefix . $scriptfix . $prefix . '<[^>]+>|&[a-zA-Z#][a-z0-9]+;' . $marker . $htmlfix . '/s';
 266
 267                 $matches = preg_split($reg, $text, -1, PREG_SPLIT_OFFSET_CAPTURE);
 268
 269                 $m = array_shift($matches);
 270
 271                 $ret = $this->translate($m[0], $toVariant);
 272                 $mstart = $m[1]+strlen($m[0]);
 273
 274                 // enable convertsion of '<img alt="xxxx" ... ' or '<span title="xxxx" ... '
 275                 $captionpattern  = '/\s(title|alt)\s*=\s*"([\s\S]*?)"/';
 276                 foreach($matches as $m) {
 277                         $mark = substr($text, $mstart, $m[1]-$mstart);
 278                         $mark = preg_replace_callback($captionpattern, array(&$this, 'captionConvert'), $mark);
 279                         $ret .= $mark;
 280                         $ret .= $this->translate($m[0], $toVariant);
 281                         $mstart = $m[1] + strlen($m[0]);
 282                 }
 283                 wfProfileOut( $fname );
 284                 return $ret;
 285         }
 286
 287         /**
 288          * Translate a string to a variant
 289          * Doesn't process markup or do any of that other stuff, for that use convert()
 290          *
 291          * @param string $text Text to convert
 292          * @param string $variant Variant language code
 293          * @return string Translated text
 294          * @private
 295          */
 296         function translate( $text, $variant ) {
 297                 wfProfileIn( __METHOD__ );
 298                 if( !$this->mTablesLoaded )
 299                         $this->loadTables();
 300                 $text = $this->mTables[$variant]->replace( $text );
 301                 wfProfileOut( __METHOD__ );
 302                 return $text;
 303         }
 304
 305         /**
 306          * convert text to all supported variants
 307          *
 308          * @param string $text the text to be converted
 309          * @return array of string
 310          * @public
 311          */
 312         function autoConvertToAllVariants($text) {
 313                 $fname="LanguageConverter::autoConvertToAllVariants";
 314                 wfProfileIn( $fname );
 315                 if( !$this->mTablesLoaded )
 316                         $this->loadTables();
 317
 318                 $ret = array();
 319                 foreach($this->mVariants as $variant) {
 320                         $ret[$variant] = $this->translate($text, $variant);
 321                 }
 322
 323                 wfProfileOut( $fname );
 324                 return $ret;
 325         }
 326
 327         /**
 328          * convert link text to all supported variants
 329          *
 330          * @param string $text the text to be converted
 331          * @return array of string
 332          * @public
 333          */
 334         function convertLinkToAllVariants($text) {
 335                 if( !$this->mTablesLoaded )
 336                         $this->loadTables();
 337
 338                 $ret = array();
 339                 $tarray = explode($this->mMarkup['begin'], $text);
 340                 $tfirst = array_shift($tarray);
 341
 342                 foreach($this->mVariants as $variant)
 343                         $ret[$variant] = $this->translate($tfirst,$variant);
 344
 345                 foreach($tarray as $txt) {
 346                         $marked = explode($this->mMarkup['end'], $txt, 2);
 347
 348                         foreach($this->mVariants as $variant){
 349                                 $ret[$variant] .= $this->mMarkup['begin'].$marked[0].$this->mMarkup['end'];
 350                                 if(array_key_exists(1, $marked))
 351                                         $ret[$variant] .= $this->translate($marked[1],$variant);
 352                         }
 353
 354                 }
 355
 356                 return $ret;
 357         }
 358
 359         /**
 360          * prepare manual conversion table
 361          * @private
 362          */
 363         function prepareManualConv( $convRule ){
 364                 // use syntax -{T|zh:TitleZh;zh-tw:TitleTw}- for custom conversion in title
 365                 $title = $convRule->getTitle();
 366                 if( $title ){
 367                         $this->mTitleFromFlag = true;
 368                         $this->mTitleDisplay =  $title;
 369                 }
 370
 371                 //apply manual conversion table to global table
 372                 $convTable = $convRule->getConvTable();
 373                 $action = $convRule->getRulesAction();
 374                 foreach( $convTable as $v => $t ) {
 375                         if( !in_array( $v, $this->mVariants ) )continue;
 376                         if( $action=="add" ) {
 377                                 foreach( $t as $from => $to ) {
 378                                         // to ensure that $from and $to not be left blank
 379                                         // so $this->translate() could always return a string
 380                                         if ( $from || $to )
 381                                                 // more efficient than array_merge(), about 2.5 times.
 382                                                 $this->mManualAddTables[$v][$from] = $to;
 383                                 }
 384                         }
 385                         elseif ( $action == "remove" ) {
 386                                 foreach ( $t as $from=>$to ) {
 387                                         if ( $from || $to )
 388                                                 $this->mManualRemoveTables[$v][$from] = $to;
 389                                 }
 390                         }
 391                 }
 392         }
 393
 394         /**
 395          * apply manual conversion from $this->mManualAddTables and $this->mManualRemoveTables
 396          * @private
 397          */
 398         function applyManualConv(){
 399                 //apply manual conversion table to global table
 400                 foreach($this->mVariants as $v) {
 401                         if (count($this->mManualAddTables[$v]) > 0) {
 402                                 $this->mTables[$v]->mergeArray($this->mManualAddTables[$v]);
 403                         }
 404                         if (count($this->mManualRemoveTables[$v]) > 0)
 405                                 $this->mTables[$v]->removeArray($this->mManualRemoveTables[$v]);
 406                 }
 407         }
 408
 409         /**
 410          * Convert text using a parser object for context
 411          * @public
 412          */
 413         function parserConvert( $text, &$parser ) {
 414                 global $wgDisableLangConversion;
 415                 /* don't do anything if this is the conversion table */
 416                 if ( $parser->getTitle()->getNamespace() == NS_MEDIAWIKI &&
 417                                  strpos($parser->mTitle->getText(), "Conversiontable") !== false )
 418                 {
 419                         return $text;
 420                 }
 421
 422                 if ( $wgDisableLangConversion )
 423                         return $text;
 424
 425                 $text = $this->convert( $text );
 426
 427                 if ( $this->mTitleFromFlag )
 428                         $parser->mOutput->setTitleText( $this->mTitleDisplay );
 429                 return $text;
 430         }
 431
 432         /**
 433          * convert namespace
 434          * @param string $title the title included namespace
 435          * @return array of string
 436          * @private
 437          */
 438         function convertNamespace( $title, $variant ) {
 439                 $splittitle = explode( ':', $title );
 440                 if (count($splittitle) < 2)
 441                         return $title;
 442                 if ( isset( $this->mNamespaceTables[$variant][$splittitle[0]] ) )
 443                         $splittitle[0] = $this->mNamespaceTables[$variant][$splittitle[0]];
 444                 $ret = implode(':', $splittitle );
 445                 return $ret;
 446         }
 447
 448         /**
 449          * convert title
 450          * @private
 451          */
 452         function convertTitle( $text, $variant ){
 453                 global $wgDisableTitleConversion, $wgUser;
 454
 455                 // check for global param and __NOTC__ tag
 456                 if( $wgDisableTitleConversion || !$this->mDoTitleConvert || $wgUser->getOption('noconvertlink') == 1 ) {
 457                         $this->mTitleDisplay = $text;
 458                         return $text;
 459                 }
 460
 461                 // use the title from the T flag if any
 462                 if( $this->mTitleFromFlag ){
 463                         $this->mTitleFromFlag = false;
 464                         return $this->mTitleDisplay;
 465                 }
 466
 467                 global $wgRequest;
 468                 $isredir = $wgRequest->getText( 'redirect', 'yes' );
 469                 $action = $wgRequest->getText( 'action' );
 470                 $linkconvert = $wgRequest->getText( 'linkconvert', 'yes' );
 471                 if ( $isredir == 'no' || $action == 'edit' || $action == 'submit' || $linkconvert == 'no' ) {
 472                         return $text;
 473                 } else {
 474                         $text = $this->convertNamespace( $text, $variant );
 475                         $this->mTitleDisplay = $this->convert( $text );
 476                         return $this->mTitleDisplay;
 477                 }
 478         }
 479
 480         /**
 481          * convert text to different variants of a language. the automatic
 482          * conversion is done in autoConvert(). here we parse the text
 483          * marked with -{}-, which specifies special conversions of the
 484          * text that can not be accomplished in autoConvert()
 485          *
 486          * syntax of the markup:
 487          * -{code1:text1;code2:text2;...}-  or
 488          * -{flags|code1:text1;code2:text2;...}-  or
 489          * -{text}- in which case no conversion should take place for text
 490          *
 491          * @param string $text text to be converted
 492          * @param bool $isTitle whether this conversion is for the article title
 493          * @return string converted text
 494          * @public
 495          */
 496         function convert( $text, $isTitle = false ) {
 497
 498                 $mw =& MagicWord::get( 'notitleconvert' );
 499                 if( $mw->matchAndRemove( $text ) )
 500                         $this->mDoTitleConvert = false;
 501                 $mw =& MagicWord::get( 'nocontentconvert' );
 502                 if( $mw->matchAndRemove( $text ) ) {
 503                         $this->mDoContentConvert = false;
 504                 }
 505
 506                 // no conversion if redirecting
 507                 $mw =& MagicWord::get( 'redirect' );
 508                 if( $mw->matchStart( $text ) )
 509                         return $text;
 510
 511                 $plang = $this->getPreferredVariant();
 512
 513                 // for title convertion
 514                 if ( $isTitle ) return $this->convertTitle( $text, $plang );
 515
 516                 $tarray = StringUtils::explode( $this->mMarkup['end'], $text );
 517                 $text = '';
 518
 519                 $marks = array();
 520                 foreach ( $tarray as $txt ) {
 521                         $marked = explode( $this->mMarkup['begin'], $txt, 2 );
 522                         if ( array_key_exists( 1, $marked ) ) {
 523                                 $crule = new ConverterRule($marked[1], $this);
 524                                 $crule->parse( $plang );
 525                                 $marked[1] = $crule->getDisplay();
 526                                 $this->prepareManualConv( $crule );
 527                         }
 528                         else
 529                                 $marked[0] .= $this->mMarkup['end'];
 530                         array_push( $marks, $marked );
 531                 }
 532                 $this->applyManualConv();
 533                 foreach ( $marks as $marked ) {
 534                         if( $this->mDoContentConvert )
 535                                 $text .= $this->autoConvert( $marked[0], $plang );
 536                         else
 537                                 $text .= $marked[0];
 538                         if( array_key_exists( 1, $marked ) )
 539                                 $text .= $marked[1];
 540                 }
 541                 // Remove the last delimiter (wasn't real)
 542                 $text = substr( $text, 0, -strlen( $this->mMarkup['end'] ) );
 543
 544                 return $text;
 545         }
 546
 547         /**
 548          * if a language supports multiple variants, it is
 549          * possible that non-existing link in one variant
 550          * actually exists in another variant. this function
 551          * tries to find it. See e.g. LanguageZh.php
 552          *
 553          * @param string $link the name of the link
 554          * @param mixed $nt the title object of the link
 555          * @param boolean $ignoreOtherCond: to disable other conditions when
 556          *      we need to transclude a template or update a category's link
 557          * @return null the input parameters may be modified upon return
 558          * @public
 559          */
 560         function findVariantLink( &$link, &$nt, $ignoreOtherCond = false ) {
 561                 global $wgDisableLangConversion, $wgDisableTitleConversion, $wgRequest, $wgUser;
 562                 $isredir = $wgRequest->getText( 'redirect', 'yes' );
 563                 $action = $wgRequest->getText( 'action' );
 564                 $linkconvert = $wgRequest->getText( 'linkconvert', 'yes' );
 565                 $disableLinkConversion = $wgDisableLangConversion || $wgDisableTitleConversion;
 566                 $linkBatch = new LinkBatch();
 567
 568                 $ns=NS_MAIN;
 569
 570                 if ( $disableLinkConversion || ( !$ignoreOtherCond && ( $isredir == 'no' || $action == 'edit'
 571                         || $action == 'submit' || $linkconvert == 'no' || $wgUser->getOption('noconvertlink') == 1 ) ) )
 572                         return;
 573
 574                 if(is_object($nt))
 575                         $ns = $nt->getNamespace();
 576
 577                 $variants = $this->autoConvertToAllVariants($link);
 578                 if($variants == false) //give up
 579                         return;
 580
 581                 $titles = array();
 582
 583                 foreach( $variants as $v ) {
 584                         if($v != $link){
 585                                 $varnt = Title::newFromText( $v, $ns );
 586                                 if(!is_null($varnt)){
 587                                         $linkBatch->addObj($varnt);
 588                                         $titles[]=$varnt;
 589                                 }
 590                         }
 591                 }
 592
 593                 // fetch all variants in single query
 594                 $linkBatch->execute();
 595
 596                 foreach( $titles as $varnt ) {
 597                         if( $varnt->getArticleID() > 0 ) {
 598                                 $nt = $varnt;
 599                                 $link = $varnt->getText();
 600                                 break;
 601                         }
 602                 }
 603         }
 604
 605     /**
 606          * returns language specific hash options
 607          *
 608          * @public
 609          */
 610         function getExtraHashOptions() {
 611                 $variant = $this->getPreferredVariant();
 612                 return '!' . $variant ;
 613         }
 614
 615     /**
 616          * get title text as defined in the body of the article text
 617          *
 618          * @public
 619          */
 620         function getParsedTitle() {
 621                 return $this->mTitleDisplay;
 622         }
 623
 624         /**
 625          * a write lock to the cache
 626          *
 627          * @private
 628          */
 629         function lockCache() {
 630                 global $wgMemc;
 631                 $success = false;
 632                 for($i=0; $i<30; $i++) {
 633                         if($success = $wgMemc->add($this->mCacheKey . "lock", 1, 10))
 634                                 break;
 635                         sleep(1);
 636                 }
 637                 return $success;
 638         }
 639
 640         /**
 641          * unlock cache
 642          *
 643          * @private
 644          */
 645         function unlockCache() {
 646                 global $wgMemc;
 647                 $wgMemc->delete($this->mCacheKey . "lock");
 648         }
 649
 650
 651         /**
 652          * Load default conversion tables
 653          * This method must be implemented in derived class
 654          *
 655          * @private
 656          */
 657         function loadDefaultTables() {
 658                 $name = get_class($this);
 659                 wfDie("Must implement loadDefaultTables() method in class $name");
 660         }
 661
 662         /**
 663          * load conversion tables either from the cache or the disk
 664          * @private
 665          */
 666         function loadTables($fromcache=true) {
 667                 global $wgMemc;
 668                 if( $this->mTablesLoaded )
 669                         return;
 670                 wfProfileIn( __METHOD__ );
 671                 $this->mTablesLoaded = true;
 672                 $this->mTables = false;
 673                 if($fromcache) {
 674                         wfProfileIn( __METHOD__.'-cache' );
 675                         $this->mTables = $wgMemc->get( $this->mCacheKey );
 676                         wfProfileOut( __METHOD__.'-cache' );
 677                 }
 678                 if ( !$this->mTables || !isset( $this->mTables[self::CACHE_VERSION_KEY] ) ) {
 679                         wfProfileIn( __METHOD__.'-recache' );
 680                         // not in cache, or we need a fresh reload.
 681                         // we will first load the default tables
 682                         // then update them using things in MediaWiki:Zhconversiontable/*
 683                         $this->loadDefaultTables();
 684                         foreach($this->mVariants as $var) {
 685                                 $cached = $this->parseCachedTable($var);
 686                                 $this->mTables[$var]->mergeArray($cached);
 687                         }
 688
 689                         $this->postLoadTables();
 690                         $this->mTables[self::CACHE_VERSION_KEY] = true;
 691
 692                         if($this->lockCache()) {
 693                                 $wgMemc->set($this->mCacheKey, $this->mTables, 43200);
 694                                 $this->unlockCache();
 695                         }
 696                         wfProfileOut( __METHOD__.'-recache' );
 697                 }
 698                 wfProfileOut( __METHOD__ );
 699         }
 700
 701     /**
 702          * Hook for post processig after conversion tables are loaded
 703          *
 704          */
 705         function postLoadTables() {}
 706
 707     /**
 708          * Reload the conversion tables
 709          *
 710          * @private
 711          */
 712         function reloadTables() {
 713                 if($this->mTables)
 714                         unset($this->mTables);
 715                 $this->mTablesLoaded = false;
 716                 $this->loadTables(false);
 717         }
 718
 719
 720         /**
 721          * parse the conversion table stored in the cache
 722          *
 723          * the tables should be in blocks of the following form:
 724          *              -{
 725          *                      word => word ;
 726          *                      word => word ;
 727          *                      ...
 728          *              }-
 729          *
 730          *      to make the tables more manageable, subpages are allowed
 731          *      and will be parsed recursively if $recursive=true
 732          *
 733          */
 734         function parseCachedTable($code, $subpage='', $recursive=true) {
 735                 global $wgMessageCache;
 736                 static $parsed = array();
 737
 738                 if(!is_object($wgMessageCache))
 739                         return array();
 740
 741                 $key = 'Conversiontable/'.$code;
 742                 if($subpage)
 743                         $key .= '/' . $subpage;
 744
 745                 if(array_key_exists($key, $parsed))
 746                         return array();
 747
 748                 if ( strpos( $code, '/' ) === false ) {
 749                         $txt = $wgMessageCache->get( 'Conversiontable', true, $code );
 750                 } else {
 751                         $title = Title::makeTitleSafe( NS_MEDIAWIKI, "Conversiontable/$code" );
 752                         if ( $title && $title->exists() ) {
 753                                 $article = new Article( $title );
 754                                 $txt = $article->getContents();
 755                         } else {
 756                                 $txt = '';
 757                         }
 758                 }
 759
 760                 // get all subpage links of the form
 761                 // [[MediaWiki:conversiontable/zh-xx/...|...]]
 762                 $linkhead = $this->mLangObj->getNsText(NS_MEDIAWIKI) . ':Conversiontable';
 763                 $subs = explode('[[', $txt);
 764                 $sublinks = array();
 765                 foreach( $subs as $sub ) {
 766                         $link = explode(']]', $sub, 2);
 767                         if(count($link) != 2)
 768                                 continue;
 769                         $b = explode('|', $link[0]);
 770                         $b = explode('/', trim($b[0]), 3);
 771                         if(count($b)==3)
 772                                 $sublink = $b[2];
 773                         else
 774                                 $sublink = '';
 775
 776                         if($b[0] == $linkhead && $b[1] == $code) {
 777                                 $sublinks[] = $sublink;
 778                         }
 779                 }
 780
 781
 782                 // parse the mappings in this page
 783                 $blocks = explode($this->mMarkup['begin'], $txt);
 784                 array_shift($blocks);
 785                 $ret = array();
 786                 foreach($blocks as $block) {
 787                         $mappings = explode($this->mMarkup['end'], $block, 2);
 788                         $stripped = str_replace(array("'", '"', '*','#'), '', $mappings[0]);
 789                         $table = explode( ';', $stripped );
 790                         foreach( $table as $t ) {
 791                                 $m = explode( '=>', $t );
 792                                 if( count( $m ) != 2)
 793                                         continue;
 794                                 // trim any trailling comments starting with '//'
 795                                 $tt = explode('//', $m[1], 2);
 796                                 $ret[trim($m[0])] = trim($tt[0]);
 797                         }
 798                 }
 799                 $parsed[$key] = true;
 800
 801
 802                 // recursively parse the subpages
 803                 if($recursive) {
 804                         foreach($sublinks as $link) {
 805                                 $s = $this->parseCachedTable($code, $link, $recursive);
 806                                 $ret = array_merge($ret, $s);
 807                         }
 808                 }
 809
 810                 if ($this->mUcfirst) {
 811                         foreach ($ret as $k => $v) {
 812                                 $ret[Language::ucfirst($k)] = Language::ucfirst($v);
 813                         }
 814                 }
 815                 return $ret;
 816         }
 817
 818         /**
 819          * Enclose a string with the "no conversion" tag. This is used by
 820          * various functions in the Parser
 821          *
 822          * @param string $text text to be tagged for no conversion
 823          * @return string the tagged text
 824          * @public
 825          */
 826         function markNoConversion($text, $noParse=false) {
 827                 # don't mark if already marked
 828                 if(strpos($text, $this->mMarkup['begin']) ||
 829                    strpos($text, $this->mMarkup['end']))
 830                         return $text;
 831
 832                 $ret = $this->mMarkup['begin'] .'R|'. $text . $this->mMarkup['end'];
 833                 return $ret;
 834         }
 835
 836         /**
 837          * convert the sorting key for category links. this should make different
 838          * keys that are variants of each other map to the same key
 839          */
 840         function convertCategoryKey( $key ) {
 841                 return $key;
 842         }
 843         /**
 844          * hook to refresh the cache of conversion tables when
 845          * MediaWiki:conversiontable* is updated
 846          * @private
 847          */
 848         function OnArticleSaveComplete($article, $user, $text, $summary, $isminor, $iswatch, $section, $flags, $revision) {
 849                 $titleobj = $article->getTitle();
 850                 if($titleobj->getNamespace() == NS_MEDIAWIKI) {
 851                         $title = $titleobj->getDBkey();
 852                         $t = explode('/', $title, 3);
 853                         $c = count($t);
 854                         if( $c > 1 && $t[0] == 'Conversiontable' ) {
 855                                 if(in_array($t[1], $this->mVariants)) {
 856                                         $this->reloadTables();
 857                                 }
 858                         }
 859                 }
 860                 return true;
 861         }
 862
 863         /**
 864          * Armour rendered math against conversion
 865          * Wrap math into rawoutput -{R| math }- syntax
 866          * @public
 867          */
 868         function armourMath($text){
 869                 $ret = $this->mMarkup['begin'] . 'R|' . $text . $this->mMarkup['end'];
 870                 return $ret;
 871         }
 872 }
 873
 874 /**
 875  * parser for rules of language conversion , parse rules in -{ }- tag
 876  * @ingroup Language
 877  * @author fdcn <fdcn64@gmail.com>, PhiLiP <philip.npc@gmail.com>
 878  */
 879 class ConverterRule {
 880         var $mText; // original text in -{text}-
 881         var $mConverter; // LanguageConverter object
 882         var $mManualCodeError='<strong class="error">code error!</strong>';
 883         var $mRuleDisplay = '',$mRuleTitle=false;
 884         var $mRules = '';// string : the text of the rules
 885         var $mRulesAction = 'none';
 886         var $mFlags = array();
 887         var $mConvTable = array();
 888         var $mBidtable = array();// array of the translation in each variant
 889         var $mUnidtable = array();// array of the translation in each variant
 890
 891         /**
 892          * Constructor
 893          *
 894          * @param string $text the text between -{ and }-
 895          * @param object $converter a  LanguageConverter object
 896          * @access public
 897          */
 898         function __construct($text,$converter){
 899                 $this->mText = $text;
 900                 $this->mConverter=$converter;
 901                 foreach($converter->mVariants as $v){
 902                         $this->mConvTable[$v]=array();
 903                 }
 904         }
 905
 906         /**
 907          * check if variants array in convert array
 908          *
 909          * @param string $variant Variant language code
 910          * @return string Translated text
 911          * @public
 912          */
 913         function getTextInBidtable($variants){
 914                 if(is_string($variants)){ $variants=array($variants); }
 915                 if(!is_array($variants)) return false;
 916                 foreach ($variants as $variant){
 917                         if(array_key_exists($variant, $this->mBidtable)){
 918                                 return $this->mBidtable[$variant];
 919                         }
 920                 }
 921                 return false;
 922         }
 923
 924         /**
 925          * Parse flags with syntax -{FLAG| ... }-
 926          * @private
 927          */
 928         function parseFlags(){
 929                 $text = $this->mText;
 930                 if(strlen($text) < 2 ) {
 931                         $this->mFlags = array( 'R' );
 932                         $this->mRules = $text;
 933                         return;
 934                 }
 935
 936                 $flags = array();
 937                 $markup = $this->mConverter->mMarkup;
 938                 $validFlags = $this->mConverter->mFlags;
 939                 $variants = $this->mConverter->mVariants;
 940
 941                 $tt = explode($markup['flagsep'], $text, 2);
 942                 if(count($tt) == 2) {
 943                         $f = explode($markup['varsep'], $tt[0]);
 944                         foreach($f as $ff) {
 945                                 $ff = trim($ff);
 946                                 if(array_key_exists($ff, $validFlags) &&
 947                                                         !in_array($validFlags[$ff], $flags))
 948                                         $flags[] = $validFlags[$ff];
 949                         }
 950                         $rules = $tt[1];
 951                 } else {
 952                         $rules = $text;
 953                 }
 954
 955                 //check flags
 956                 if( in_array('R',$flags) ){
 957                         $flags = array('R');// remove other flags
 958                 } elseif ( in_array('N',$flags) ){
 959                         $flags = array('N');// remove other flags
 960                 } elseif ( in_array('-',$flags) ){
 961                         $flags = array('-');// remove other flags
 962                 } elseif (count($flags)==1 && $flags[0]=='T'){
 963                         $flags[]='H';
 964                 } elseif ( in_array('H',$flags) ){
 965                         // replace A flag, and remove other flags except T
 966                         $temp=array('+','H');
 967                         if(in_array('T',$flags)) $temp[] = 'T';
 968                         if(in_array('D',$flags)) $temp[] = 'D';
 969                         $flags = $temp;
 970                 } else {
 971                         if ( in_array('A',$flags) ) {
 972                                 $flags[]='+';
 973                                 $flags[]='S';
 974                         }
 975                         if ( in_array('D',$flags) )
 976                                 $flags=array_diff($flags,array('S'));
 977                         $flags_temp = array();
 978                         foreach ($variants as $variant) {
 979                                 // try to find flags like "zh-hans", "zh-hant"
 980                                 // allow syntaxes like "-{zh-hans;zh-hant|XXXX}-"
 981                                 if ( in_array($variant, $flags) )
 982                                         $flags_temp[] = $variant;
 983                         }
 984                         if ( count($flags_temp) == 0 )
 985                                 $flags = $flags_temp;
 986                 }
 987                 if ( count($flags) == 0 )
 988                         $flags = array('S');
 989                 $this->mRules=$rules;
 990                 $this->mFlags=$flags;
 991         }
 992
 993         /**
 994          * generate conversion table
 995          * @private
 996          */
 997         function parseRules() {
 998                 $rules = $this->mRules;
 999                 $flags = $this->mFlags;
1000                 $bidtable = array();
1001                 $unidtable = array();
1002                 $markup = $this->mConverter->mMarkup;
1003                 $variants = $this->mConverter->mVariants;
1004
1005                 // varsep_pattern for preg_split:
1006                 // text should be splited by ";" only if a valid variant
1007                 // name exist after the markup, for example:
1008                 //  -{zh-hans:<span style="font-size:120%;">xxx</span>;zh-hant:<span style="font-size:120%;">yyy</span>;}-
1009                 // we should split it as:
1010                 //  array(
1011                 //        [0] => 'zh-hans:<span style="font-size:120%;">xxx</span>'
1012                 //        [1] => 'zh-hant:<span style="font-size:120%;">yyy</span>'
1013                 //        [2] => ''
1014                 //       )
1015                 $varsep_pattern = '/' . $markup['varsep'] . '\s*' . '(?=';
1016                 foreach( $variants as $variant ) {
1017                         $varsep_pattern .= $variant . '\s*' . $markup['codesep'] . '|'; // zh-hans:xxx;zh-hant:yyy
1018                         $varsep_pattern .= '[^;]*?' . $markup['unidsep'] . '\s*' . $variant
1019                                                         . '\s*' . $markup['codesep'] . '|'; // xxx=>zh-hans:yyy; xxx=>zh-hant:zzz
1020                 }
1021                 $varsep_pattern .= '\s*$)/';
1022
1023                 $choice = preg_split($varsep_pattern, $rules);
1024
1025                 foreach( $choice as $c ) {
1026                         $v  = explode($markup['codesep'], $c, 2);
1027                         if( count($v) != 2 )
1028                                 continue;// syntax error, skip
1029                         $to = trim($v[1]);
1030                         $v  = trim($v[0]);
1031                         $u  = explode($markup['unidsep'], $v);
1032                         // if $to is empty, strtr() could return a wrong result
1033                         if( count($u) == 1 && $to && in_array( $v, $variants ) ) {
1034                                 $bidtable[$v] = $to;
1035                         } else if(count($u) == 2){
1036                                 $from = trim($u[0]);
1037                                 $v    = trim($u[1]);
1038                                 if( array_key_exists( $v, $unidtable ) && !is_array( $unidtable[$v] )
1039                                         && $to && in_array( $v, $variants ) )
1040                                         $unidtable[$v] = array( $from=>$to );
1041                                 elseif ( $to && in_array( $v, $variants ) )
1042                                         $unidtable[$v][$from] = $to;
1043                         }
1044                         // syntax error, pass
1045                         if ( !array_key_exists( $v, $this->mConverter->mVariantNames ) ){
1046                                 $bidtable = array();
1047                                 $unidtable = array();
1048                                 break;
1049                         }
1050                 }
1051                 $this->mBidtable = $bidtable;
1052                 $this->mUnidtable = $unidtable;
1053         }
1054
1055         /**
1056          * @private
1057          */
1058         function getRulesDesc(){
1059                 $codesep = $this->mConverter->mDescCodeSep;
1060                 $varsep = $this->mConverter->mDescVarSep;
1061                 $text='';
1062                 foreach($this->mBidtable as $k => $v)
1063                         $text .= $this->mConverter->mVariantNames[$k]."$codesep$v$varsep";
1064                 foreach($this->mUnidtable as $k => $a)
1065                         foreach($a as $from=>$to)
1066                                 $text.=$from.'⇒'.$this->mConverter->mVariantNames[$k]."$codesep$to$varsep";
1067                 return $text;
1068         }
1069
1070         /**
1071          *  Parse rules conversion
1072          * @private
1073          */
1074         function getRuleConvertedStr($variant,$doConvert){
1075                 $bidtable = $this->mBidtable;
1076                 $unidtable = $this->mUnidtable;
1077
1078                 if( count($bidtable) + count($unidtable) == 0 ){
1079                         return $this->mRules;
1080                 } elseif ($doConvert){// the text converted
1081                         // display current variant in bidirectional array
1082                         $disp = $this->getTextInBidtable($variant);
1083                         // or display current variant in fallbacks
1084                         if(!$disp)
1085                                 $disp = $this->getTextInBidtable(
1086                                                 $this->mConverter->getVariantFallbacks($variant));
1087                         // or display current variant in unidirectional array
1088                         if(!$disp && array_key_exists($variant,$unidtable)){
1089                                 $disp = array_values($unidtable[$variant]);
1090                                 $disp = $disp[0];
1091                         }
1092                         // or display frist text under disable manual convert
1093                         if(!$disp && $this->mConverter->mManualLevel[$variant]=='disable') {
1094                                 if(count($bidtable)>0){
1095                                         $disp = array_values($bidtable);
1096                                         $disp = $disp[0];
1097                                 } else {
1098                                         $disp = array_values($unidtable);
1099                                         $disp = array_values($disp[0]);
1100                                         $disp = $disp[0];
1101                                 }
1102                         }
1103                         return $disp;
1104                 } else {// no convert
1105                         return $this->mRules;
1106                 }
1107         }
1108
1109         /**
1110          * generate conversion table for all text
1111          * @private
1112          */
1113         function generateConvTable(){
1114                 $flags = $this->mFlags;
1115                 $bidtable = $this->mBidtable;
1116                 $unidtable = $this->mUnidtable;
1117                 $manLevel = $this->mConverter->mManualLevel;
1118
1119                 $vmarked=array();
1120                 foreach($this->mConverter->mVariants as $v) {
1121                         /* for bidirectional array
1122                                 fill in the missing variants, if any,
1123                                 with fallbacks */
1124                         if(!array_key_exists($v, $bidtable)) {
1125                                 $variantFallbacks = $this->mConverter->getVariantFallbacks($v);
1126                                 $vf = $this->getTextInBidtable($variantFallbacks);
1127                                 if($vf) $bidtable[$v] = $vf;
1128                         }
1129
1130                         if(array_key_exists($v,$bidtable)){
1131                                 foreach($vmarked as $vo){
1132                                         // use syntax: -{A|zh:WordZh;zh-tw:WordTw}-
1133                                         // or -{H|zh:WordZh;zh-tw:WordTw}- or -{-|zh:WordZh;zh-tw:WordTw}-
1134                                         // to introduce a custom mapping between
1135                                         // words WordZh and WordTw in the whole text
1136                                         if($manLevel[$v]=='bidirectional'){
1137                                                 $this->mConvTable[$v][$bidtable[$vo]]=$bidtable[$v];
1138                                         }
1139                                         if($manLevel[$vo]=='bidirectional'){
1140                                                 $this->mConvTable[$vo][$bidtable[$v]]=$bidtable[$vo];
1141                                         }
1142                                 }
1143                                 $vmarked[]=$v;
1144                         }
1145                         /*for unidirectional array
1146                                 fill to convert tables */
1147                         $allow_unid = $manLevel[$v]=='bidirectional'
1148                                         || $manLevel[$v]=='unidirectional';
1149                         if($allow_unid && array_key_exists($v,$unidtable)){
1150                                 $ct=$this->mConvTable[$v];
1151                                 $this->mConvTable[$v] = array_merge($ct,$unidtable[$v]);
1152                         }
1153                 }
1154         }
1155
1156         /**
1157          * Parse rules and flags
1158          * @public
1159          */
1160         function parse($variant){
1161                 if(!$variant)
1162                         $variant = $this->mConverter->getPreferredVariant();
1163
1164                 $variants = $this->mConverter->mVariants;
1165                 $this->parseFlags();
1166                 $flags = $this->mFlags;
1167
1168                 // convert to specified variant
1169                 // syntax: -{zh-hans;zh-hant[;...]|<text to convert>}-
1170                 if( count( array_diff( $flags, $variants ) ) == 0 and count( $flags ) != 0 ) {
1171                         if ( in_array( $variant, $flags ) ) // check if current variant in flags
1172                                 // then convert <text to convert> to current language
1173                                 $this->mRules = $this->mConverter->autoConvert( $this->mRules, $variant );
1174                         else { // if current variant no in flags,
1175                                    // then we check its fallback variants.
1176                                 $variantFallbacks = $this->mConverter->getVariantFallbacks($variant);
1177                                 foreach ( $variantFallbacks as $variantFallback ) {
1178                                         // if current variant's fallback exist in flags
1179                                         if ( in_array( $variantFallback, $flags ) ) {
1180                                                 // then convert <text to convert> to fallback language
1181                                                 $this->mRules = $this->mConverter->autoConvert( $this->mRules, $variantFallback );
1182                                                 break;
1183                                         }
1184                                 }
1185                         }
1186                         $this->mFlags = $flags = array('R');
1187                 }
1188
1189                 if( !in_array( 'R', $flags ) || !in_array( 'N', $flags ) ) {
1190                         // decode => HTML entities modified by Sanitizer::removeHTMLtags
1191                         $this->mRules = str_replace('=&gt;','=>',$this->mRules);
1192
1193                         $this->parseRules();
1194                 }
1195                 $rules = $this->mRules;
1196
1197                 if( count( $this->mBidtable ) == 0 && count( $this->mUnidtable ) == 0 ){
1198                         if(in_array('+',$flags) || in_array('-',$flags))
1199                                 // fill all variants if text in -{A/H/-|text} without rules
1200                                 foreach($this->mConverter->mVariants as $v)
1201                                         $this->mBidtable[$v] = $rules;
1202                         elseif (!in_array('N',$flags) && !in_array('T',$flags) )
1203                                 $this->mFlags = $flags = array('R');
1204                 }
1205
1206                 if( in_array('R',$flags) ) {
1207                         // if we don't do content convert, still strip the -{}- tags
1208                         $this->mRuleDisplay = $rules;
1209                 } elseif ( in_array('N',$flags) ){
1210                         // proces N flag: output current variant name
1211                         $this->mRuleDisplay = $this->mConverter->mVariantNames[trim($rules)];
1212                 } elseif ( in_array('D',$flags) ){
1213                         // proces D flag: output rules description
1214                         $this->mRuleDisplay = $this->getRulesDesc();
1215                 } elseif ( in_array('H',$flags) || in_array('-',$flags) ) {
1216                         // proces H,- flag or T only: output nothing
1217                         $this->mRuleDisplay = '';
1218                 } elseif ( in_array('S',$flags) ){
1219                         $this->mRuleDisplay = $this->getRuleConvertedStr($variant,
1220                                                         $this->mConverter->mDoContentConvert);
1221                 } else {
1222                         $this->mRuleDisplay= $this->mManualCodeError;
1223                 }
1224                 // proces T flag
1225                 if ( in_array('T',$flags) ) {
1226                         $this->mRuleTitle = $this->getRuleConvertedStr($variant,
1227                                                         $this->mConverter->mDoTitleConvert);
1228                 }
1229
1230                 if (in_array('-', $flags))
1231                         $this->mRulesAction='remove';
1232                 if (in_array('+', $flags))
1233                         $this->mRulesAction='add';
1234
1235                 $this->generateConvTable();
1236         }
1237
1238         /**
1239          * @public
1240          */
1241         function hasRules(){
1242                 // TODO:
1243         }
1244
1245         /**
1246          * get display text on markup -{...}-
1247          * @public
1248          */
1249         function getDisplay(){
1250                 return $this->mRuleDisplay;
1251         }
1252         /**
1253          * get converted title
1254          * @public
1255          */
1256         function getTitle(){
1257                 return $this->mRuleTitle;
1258         }
1259
1260         /**
1261          * return how deal with conversion rules
1262          * @public
1263          */
1264         function getRulesAction(){
1265                 return $this->mRulesAction;
1266         }
1267
1268         /**
1269          * get conversion table ( bidirectional and unidirectional conversion table )
1270          * @public
1271          */
1272         function getConvTable(){
1273                 return $this->mConvTable;
1274         }
1275
1276         /**
1277          * get conversion rules string
1278          * @public
1279          */
1280         function getRules(){
1281                 return $this->mRules;
1282         }
1283
1284         /**
1285          * get conversion flags
1286          * @public
1287          */
1288         function getFlags(){
1289                 return $this->mFlags;
1290         }
1291 }