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 $mNamespaceTables;
  25         var $mDoTitleConvert=true, $mDoContentConvert=true;
  26         var $mManualLevel; // 'bidirectional' 'unidirectional' 'disable' for each variants
  27         var $mTitleFromFlag = false;
  28         var $mCacheKey;
  29         var $mLangObj;
  30         var $mMarkup;
  31         var $mFlags;
  32         var $mDescCodeSep = ':',$mDescVarSep = ';';
  33         var $mUcfirst = false;
  34         var $mTitleOriginal = '';
  35         var $mTitleDisplay = '';
  36
  37         const CACHE_VERSION_KEY = 'VERSION 6';
  38
  39         /**
  40          * Constructor
  41          *
  42          * @param string $maincode the main language code of this language
  43          * @param array $variants the supported variants of this language
  44          * @param array $variantfallback the fallback language of each variant
  45          * @param array $markup array defining the markup used for manual conversion
  46          * @param array $flags array defining the custom strings that maps to the flags
  47          * @param array $manualLevel limit for supported variants
  48          * @public
  49          */
  50         function __construct( $langobj, $maincode,
  51                                                                 $variants=array(),
  52                                                                 $variantfallbacks=array(),
  53                                                                 $markup=array(),
  54                                                                 $flags = array(),
  55                                                                 $manualLevel = array() ) {
  56                 $this->mLangObj = $langobj;
  57                 $this->mMainLanguageCode = $maincode;
  58
  59                 global $wgDisabledVariants;
  60                 $this->mVariants = array();
  61                 foreach( $variants as $variant ) {
  62                         if( !in_array( $variant, $wgDisabledVariants ) )
  63                                 $this->mVariants[] = $variant;
  64                 }
  65                 $this->mVariantFallbacks = $variantfallbacks;
  66                 global $wgLanguageNames;
  67                 $this->mVariantNames = $wgLanguageNames;
  68                 $this->mCacheKey = wfMemcKey( 'conversiontables', $maincode );
  69                 $m = array(
  70                         'begin'=>'-{',
  71                         'flagsep'=>'|',
  72                         'unidsep'=>'=>', //for unidirectional conversion
  73                         'codesep'=>':',
  74                         'varsep'=>';',
  75                         'end'=>'}-'
  76                 );
  77                 $this->mMarkup = array_merge($m, $markup);
  78                 $f = array(
  79                         // 'S' show converted text
  80                         // '+' add rules for alltext
  81                         // 'E' the gave flags is error
  82                         // these flags above are reserved for program
  83                         'A'=>'A',       // add rule for convert code (all text convert)
  84                         'T'=>'T',       // title convert
  85                         'R'=>'R',       // raw content
  86                         'D'=>'D',       // convert description (subclass implement)
  87                         '-'=>'-',       // remove convert (not implement)
  88                         'H'=>'H',       // add rule for convert code (but no display in placed code )
  89                         'N'=>'N'        // current variant name
  90                 );
  91                 $this->mFlags = array_merge($f, $flags);
  92                 foreach( $this->mVariants as $v) {
  93                         $this->mManualLevel[$v]=array_key_exists($v,$manualLevel)
  94                                                                 ?$manualLevel[$v]
  95                                                                 :'bidirectional';
  96                         $this->mNamespaceTables[$v] = array();
  97                         $this->mFlags[$v] = $v;
  98                 }
  99         }
 100
 101         /**
 102          * @public
 103          */
 104         function getVariants() {
 105                 return $this->mVariants;
 106         }
 107
 108         /**
 109          * in case some variant is not defined in the markup, we need
 110          * to have some fallback. for example, in zh, normally people
 111          * will define zh-hans and zh-hant, but less so for zh-sg or zh-hk.
 112          * when zh-sg is preferred but not defined, we will pick zh-hans
 113          * in this case. right now this is only used by zh.
 114          *
 115          * @param string $v the language code of the variant
 116          * @return string array the code of the fallback language or false if there is no fallback
 117          * @public
 118          */
 119         function getVariantFallbacks($v) {
 120                 if( isset( $this->mVariantFallbacks[$v] ) ) {
 121                         return $this->mVariantFallbacks[$v];
 122                 }
 123                 return $this->mMainLanguageCode;
 124         }
 125
 126         /**
 127          * get preferred language variants.
 128          * @param boolean $fromUser Get it from $wgUser's preferences
 129          * @return string the preferred language code
 130          * @public
 131          */
 132         function getPreferredVariant( $fromUser = true ) {
 133                 global $wgUser, $wgRequest, $wgVariantArticlePath, $wgDefaultLanguageVariant;
 134
 135                 if($this->mPreferredVariant)
 136                         return $this->mPreferredVariant;
 137
 138                 // figure out user lang without constructing wgLang to avoid infinite recursion
 139                 if( $fromUser )
 140                         $defaultUserLang = $wgUser->getOption( 'language' );
 141                 else
 142                         $defaultUserLang = $this->mMainLanguageCode;
 143                 $userLang = $wgRequest->getVal( 'uselang', $defaultUserLang );
 144                 // see if interface language is same as content, if not, prevent conversion
 145                 if( ! in_array( $userLang, $this->mVariants ) ){
 146                         $this->mPreferredVariant = $this->mMainLanguageCode; // no conversion
 147                         return $this->mPreferredVariant;
 148                 }
 149
 150                 // see if the preference is set in the request
 151                 $req = $wgRequest->getText( 'variant' );
 152                 if( in_array( $req, $this->mVariants ) ) {
 153                         $this->mPreferredVariant = $req;
 154                         return $req;
 155                 }
 156
 157                 // check the syntax /code/ArticleTitle
 158                 if($wgVariantArticlePath!=false && isset($_SERVER['SCRIPT_NAME'])){
 159                         // Note: SCRIPT_NAME probably won't hold the correct value if PHP is run as CGI
 160                         // (it will hold path to php.cgi binary), and might not exist on some very old PHP installations
 161                         $scriptBase = basename( $_SERVER['SCRIPT_NAME'] );
 162                         if(in_array($scriptBase,$this->mVariants)){
 163                                 $this->mPreferredVariant = $scriptBase;
 164                                 return $this->mPreferredVariant;
 165                         }
 166                 }
 167
 168                 // get language variant preference from logged in users
 169                 // Don't call this on stub objects because that causes infinite
 170                 // recursion during initialisation
 171                 if( $fromUser && $wgUser->isLoggedIn() )  {
 172                         $this->mPreferredVariant = $wgUser->getOption('variant');
 173                         return $this->mPreferredVariant;
 174                 }
 175
 176                 // see if default variant is globaly set
 177                 if($wgDefaultLanguageVariant != false  &&  in_array( $wgDefaultLanguageVariant, $this->mVariants )){
 178                         $this->mPreferredVariant = $wgDefaultLanguageVariant;
 179                         return $this->mPreferredVariant;
 180                 }
 181
 182                 if( !$this->mPreferredVariant ) {
 183                         // see if some supported language variant is set in the
 184                         // http header, but we don't set the mPreferredVariant
 185                         // variable in case this is called before the user's
 186                         // preference is loaded
 187                         if( array_key_exists( 'HTTP_ACCEPT_LANGUAGE', $_SERVER ) ) {
 188                                 $acceptLanguage = strtolower( $_SERVER['HTTP_ACCEPT_LANGUAGE'] );
 189
 190                                 // explode by comma
 191                                 $result = explode(',', $acceptLanguage);
 192
 193                                 $languages = array();
 194
 195                                 foreach( $result as $elem ) {
 196                                         // if $elem likes 'zh-cn;q=0.9'
 197                                         if(($posi = strpos( $elem, ';' )) !== false ) {
 198                                                 // get the real language code likes 'zh-cn'
 199                                                 $languages[] = substr( $elem, 0, $posi );
 200                                         }
 201                                         else {
 202                                                 $languages[] = $elem;
 203                                         }
 204                                 }
 205
 206                                 $fallback_languages = array();
 207                                 foreach( $languages as $language ) {
 208                                         // strip whitespace
 209                                         $language = trim( $language );
 210                                         if( in_array( $language, $this->mVariants ) ) {
 211                                                 return $language;
 212                                         }
 213                                         else {
 214                                                 // To see if there are fallbacks of current language.
 215                                                 // We record these fallback variants, and process
 216                                                 // them later.
 217                                                 $fallbacks = $this->getVariantFallbacks( $language );
 218                                                 if( is_string( $fallbacks ) )
 219                                                         $fallback_languages[] = $fallbacks;
 220                                                 elseif( is_array( $fallbacks ) )
 221                                                         $fallback_languages = array_merge( $fallback_languages, $fallbacks );
 222                                         }
 223                                 }
 224
 225                                 // process fallback languages now
 226                                 $fallback_languages = array_unique( $fallback_languages );
 227                                 foreach( $fallback_languages as $language ) {
 228                                         if( in_array( $language, $this->mVariants ) ) {
 229                                                 return $language;
 230                                         }
 231                                 }
 232                         }
 233                 }
 234
 235                 return $this->mMainLanguageCode;
 236         }
 237
 238         /**
 239          * caption convert, base on preg_replace_callback
 240          *
 241          * to convert text in "title" or "alt", like '<img alt="text" ... '
 242          * or '<span title="text" ... '
 243          *
 244          * @return string like ' alt="yyyy"' or ' title="yyyy"'
 245          * @private
 246          */
 247         function captionConvert( $matches ) {
 248                 $toVariant = $this->getPreferredVariant();
 249                 $title = $matches[1];
 250                 $text  = $matches[2];
 251                 // we convert captions except URL
 252                 if( !strpos( $text, '://' ) )
 253                         $text = $this->translate($text, $toVariant);
 254                 return " $title=\"$text\"";
 255         }
 256
 257         /**
 258          * dictionary-based conversion
 259          *
 260          * @param string $text the text to be converted
 261          * @param string $toVariant the target language code
 262          * @return string the converted text
 263          * @private
 264          */
 265         function autoConvert($text, $toVariant=false) {
 266                 $fname="LanguageConverter::autoConvert";
 267
 268                 wfProfileIn( $fname );
 269
 270                 if(!$this->mTablesLoaded)
 271                         $this->loadTables();
 272
 273                 if(!$toVariant)
 274                         $toVariant = $this->getPreferredVariant();
 275                 if(!in_array($toVariant, $this->mVariants))
 276                         return $text;
 277
 278                 /* we convert everything except:
 279                    1. html markups (anything between < and >)
 280                    2. html entities
 281                    3. place holders created by the parser
 282                 */
 283                 global $wgParser;
 284                 if (isset($wgParser) && $wgParser->UniqPrefix()!=''){
 285                         $marker = '|' . $wgParser->UniqPrefix() . '[\-a-zA-Z0-9]+';
 286                 } else
 287                         $marker = "";
 288
 289                 // this one is needed when the text is inside an html markup
 290                 $htmlfix = '|<[^>]+$|^[^<>]*>';
 291
 292                 // disable convert to variants between <code></code> tags
 293                 $codefix = '<code>.+?<\/code>|';
 294                 // disable convertsion of <script type="text/javascript"> ... </script>
 295                 $scriptfix = '<script.*?>.*?<\/script>|';
 296                 // disable conversion of <pre xxxx> ... </pre>
 297                 $prefix = '<pre.*?>.*?<\/pre>|';
 298
 299                 $reg = '/'.$codefix . $scriptfix . $prefix . '<[^>]+>|&[a-zA-Z#][a-z0-9]+;' . $marker . $htmlfix . '/s';
 300
 301                 $matches = preg_split($reg, $text, -1, PREG_SPLIT_OFFSET_CAPTURE);
 302
 303                 $m = array_shift($matches);
 304
 305                 $ret = $this->translate($m[0], $toVariant);
 306                 $mstart = $m[1]+strlen($m[0]);
 307
 308                 // enable convertsion of '<img alt="xxxx" ... ' or '<span title="xxxx" ... '
 309                 $captionpattern  = '/\s(title|alt)\s*=\s*"([\s\S]*?)"/';
 310                 foreach($matches as $m) {
 311                         $mark = substr($text, $mstart, $m[1]-$mstart);
 312                         $mark = preg_replace_callback($captionpattern, array(&$this, 'captionConvert'), $mark);
 313                         $ret .= $mark;
 314                         $ret .= $this->translate($m[0], $toVariant);
 315                         $mstart = $m[1] + strlen($m[0]);
 316                 }
 317                 wfProfileOut( $fname );
 318                 return $ret;
 319         }
 320
 321         /**
 322          * Translate a string to a variant
 323          * Doesn't process markup or do any of that other stuff, for that use convert()
 324          *
 325          * @param string $text Text to convert
 326          * @param string $variant Variant language code
 327          * @return string Translated text
 328          * @private
 329          */
 330         function translate( $text, $variant ) {
 331                 wfProfileIn( __METHOD__ );
 332                 if( !$this->mTablesLoaded )
 333                         $this->loadTables();
 334                 $text = $this->mTables[$variant]->replace( $text );
 335                 wfProfileOut( __METHOD__ );
 336                 return $text;
 337         }
 338
 339         /**
 340          * convert text to all supported variants
 341          *
 342          * @param string $text the text to be converted
 343          * @return array of string
 344          * @public
 345          */
 346         function autoConvertToAllVariants($text) {
 347                 $fname="LanguageConverter::autoConvertToAllVariants";
 348                 wfProfileIn( $fname );
 349                 if( !$this->mTablesLoaded )
 350                         $this->loadTables();
 351
 352                 $ret = array();
 353                 foreach($this->mVariants as $variant) {
 354                         $ret[$variant] = $this->translate($text, $variant);
 355                 }
 356
 357                 wfProfileOut( $fname );
 358                 return $ret;
 359         }
 360
 361         /**
 362          * convert link text to all supported variants
 363          *
 364          * @param string $text the text to be converted
 365          * @return array of string
 366          * @public
 367          */
 368         function convertLinkToAllVariants($text) {
 369                 if( !$this->mTablesLoaded )
 370                         $this->loadTables();
 371
 372                 $ret = array();
 373                 $tarray = explode($this->mMarkup['begin'], $text);
 374                 $tfirst = array_shift($tarray);
 375
 376                 foreach($this->mVariants as $variant)
 377                         $ret[$variant] = $this->translate($tfirst,$variant);
 378
 379                 foreach($tarray as $txt) {
 380                         $marked = explode($this->mMarkup['end'], $txt, 2);
 381
 382                         foreach($this->mVariants as $variant){
 383                                 $ret[$variant] .= $this->mMarkup['begin'].$marked[0].$this->mMarkup['end'];
 384                                 if(array_key_exists(1, $marked))
 385                                         $ret[$variant] .= $this->translate($marked[1],$variant);
 386                         }
 387
 388                 }
 389
 390                 return $ret;
 391         }
 392
 393         /**
 394          * prepare manual conversion table
 395          * @private
 396          */
 397         function applyManualConv( $convRule ){
 398                 // use syntax -{T|zh:TitleZh;zh-tw:TitleTw}- for custom conversion in title
 399                 $title = $convRule->getTitle();
 400                 if( $title ){
 401                         $this->mTitleFromFlag = true;
 402                         $this->mTitleDisplay =  $title;
 403                 }
 404
 405                 //apply manual conversion table to global table
 406                 $convTable = $convRule->getConvTable();
 407                 $action = $convRule->getRulesAction();
 408                 foreach( $convTable as $variant => $pair ) {
 409                         if( !in_array( $variant, $this->mVariants ) )continue;
 410                         if( $action=="add" ) {
 411                                 foreach( $pair as $from => $to ) {
 412                                         // to ensure that $from and $to not be left blank
 413                                         // so $this->translate() could always return a string
 414                                         if ( $from || $to )
 415                                                 // more efficient than array_merge(), about 2.5 times.
 416                                                 $this->mTables[$variant]->setPair( $from, $to );
 417                                 }
 418                         }
 419                         elseif ( $action == "remove" ) {
 420                                 $this->mTables[$variant]->removeArray( $pair );
 421                         }
 422                 }
 423         }
 424
 425         /**
 426          * Convert text using a parser object for context
 427          * @public
 428          */
 429         function parserConvert( $text, &$parser ) {
 430                 global $wgDisableLangConversion;
 431                 /* don't do anything if this is the conversion table */
 432                 if ( $parser->getTitle()->getNamespace() == NS_MEDIAWIKI &&
 433                                  strpos($parser->mTitle->getText(), "Conversiontable") !== false )
 434                 {
 435                         return $text;
 436                 }
 437
 438                 if ( $wgDisableLangConversion )
 439                         return $text;
 440
 441                 $text = $this->convert( $text );
 442
 443                 $this->convertTitle();
 444                 $parser->mOutput->setTitleText( $this->mTitleDisplay );
 445
 446                 return $text;
 447         }
 448
 449         /**
 450          * convert namespace
 451          * @param string $title the title included namespace
 452          * @return array of string
 453          * @private
 454          */
 455         function convertNamespace( $title, $variant ) {
 456                 $splittitle = explode( ':', $title );
 457                 if (count($splittitle) < 2)
 458                         return $title;
 459                 if ( isset( $this->mNamespaceTables[$variant][$splittitle[0]] ) )
 460                         $splittitle[0] = $this->mNamespaceTables[$variant][$splittitle[0]];
 461                 $ret = implode(':', $splittitle );
 462                 return $ret;
 463         }
 464
 465         /**
 466          * Pre convert title. Store the original title $this->mTitleOrginal;
 467          * store the default converted title to $this->mTitleDisplay.
 468          * @private
 469          */
 470         function preConvertTitle( $text, $variant ){
 471                 $this->mTitleOriginal = $text;
 472
 473                 $text = $this->convertNamespace( $text, $variant );
 474                 $this->mTitleDisplay = $this->convert( $text );
 475         }
 476
 477         /**
 478          * convert title
 479          * @private
 480          */
 481         function convertTitle(){
 482                 global $wgDisableTitleConversion, $wgUser, $wgRequest;
 483                 $isredir = $wgRequest->getText( 'redirect', 'yes' );
 484                 $action = $wgRequest->getText( 'action' );
 485                 $linkconvert = $wgRequest->getText( 'linkconvert', 'yes' );
 486
 487                 // check for the global variable, __NOTC__ magic word, and user setting
 488                 if( $wgDisableTitleConversion || !$this->mDoTitleConvert ||
 489                     $wgUser->getOption('noconvertlink') == 1 ) {
 490                         $this->mTitleDisplay = $this->mTitleOriginal;
 491                 }
 492
 493                 // check for GET params
 494                 elseif  ( $isredir == 'no' || $action == 'edit' || $linkconvert == 'no' ) {
 495                         $this->mTitleDisplay = $this->mTitleOriginal;
 496                 }
 497         }
 498
 499         /**
 500          * convert text to different variants of a language. the automatic
 501          * conversion is done in autoConvert(). here we parse the text
 502          * marked with -{}-, which specifies special conversions of the
 503          * text that can not be accomplished in autoConvert()
 504          *
 505          * syntax of the markup:
 506          * -{code1:text1;code2:text2;...}-  or
 507          * -{flags|code1:text1;code2:text2;...}-  or
 508          * -{text}- in which case no conversion should take place for text
 509          *
 510          * @param string $text text to be converted
 511          * @param bool $isTitle whether this conversion is for the article title
 512          * @return string converted text
 513          * @public
 514          */
 515         function convert( $text, $isTitle = false ) {
 516
 517                 $mw =& MagicWord::get( 'notitleconvert' );
 518                 if( $mw->matchAndRemove( $text ) )
 519                         $this->mDoTitleConvert = false;
 520                 $mw =& MagicWord::get( 'nocontentconvert' );
 521                 if( $mw->matchAndRemove( $text ) ) {
 522                         $this->mDoContentConvert = false;
 523                 }
 524
 525                 // no conversion if redirecting
 526                 $mw =& MagicWord::get( 'redirect' );
 527                 if( $mw->matchStart( $text ) )
 528                         return $text;
 529
 530                 $plang = $this->getPreferredVariant();
 531
 532                 // for title convertion
 533                 if ( $isTitle ) {
 534                         $this->preConvertTitle( $text, $plang );
 535                         return $text;
 536                 }
 537
 538                 $tarray = StringUtils::explode( $this->mMarkup['end'], $text );
 539                 $text = '';
 540
 541                 foreach ( $tarray as $txt ) {
 542
 543                         $marked = explode( $this->mMarkup['begin'], $txt, 2 );
 544
 545                         if( $this->mDoContentConvert )
 546                                 // Bug 19620: should convert a string immediately after a new rule added.
 547                                 $text .= $this->autoConvert( $marked[0], $plang );
 548                         else
 549                                 $text .= $marked[0];
 550
 551                         if ( array_key_exists( 1, $marked ) ) {
 552                                 $crule = new ConverterRule($marked[1], $this);
 553                                 $crule->parse( $plang );
 554                                 $text .= $crule->getDisplay();
 555                                 $this->applyManualConv( $crule );
 556                         }
 557                         else
 558                                 $text .= $this->mMarkup['end'];
 559
 560                 }
 561
 562                 // Remove the last delimiter (wasn't real)
 563                 $text = substr( $text, 0, -strlen( $this->mMarkup['end'] ) );
 564                 return $text;
 565         }
 566
 567         /**
 568          * if a language supports multiple variants, it is
 569          * possible that non-existing link in one variant
 570          * actually exists in another variant. this function
 571          * tries to find it. See e.g. LanguageZh.php
 572          *
 573          * @param string $link the name of the link
 574          * @param mixed $nt the title object of the link
 575          * @param boolean $ignoreOtherCond: to disable other conditions when
 576          *      we need to transclude a template or update a category's link
 577          * @return null the input parameters may be modified upon return
 578          * @public
 579          */
 580         function findVariantLink( &$link, &$nt, $ignoreOtherCond = false ) {
 581                 # If the article has already existed, there is no need to
 582                 # check it again, otherwise it may cause a fault.
 583                 if ( is_object( $nt ) && $nt->exists() )
 584                         return;
 585
 586                 global $wgDisableLangConversion, $wgDisableTitleConversion, $wgRequest, $wgUser;
 587                 $isredir = $wgRequest->getText( 'redirect', 'yes' );
 588                 $action = $wgRequest->getText( 'action' );
 589                 $linkconvert = $wgRequest->getText( 'linkconvert', 'yes' );
 590                 $disableLinkConversion = $wgDisableLangConversion || $wgDisableTitleConversion;
 591                 $linkBatch = new LinkBatch();
 592
 593                 $ns=NS_MAIN;
 594
 595                 if ( $disableLinkConversion || ( !$ignoreOtherCond && ( $isredir == 'no' || $action == 'edit'
 596                         || $action == 'submit' || $linkconvert == 'no' || $wgUser->getOption('noconvertlink') == 1 ) ) )
 597                         return;
 598
 599                 if ( is_object( $nt ) )
 600                         $ns = $nt->getNamespace();
 601
 602                 $variants = $this->autoConvertToAllVariants($link);
 603                 if($variants == false) //give up
 604                         return;
 605
 606                 $titles = array();
 607
 608                 foreach( $variants as $v ) {
 609                         if($v != $link){
 610                                 $varnt = Title::newFromText( $v, $ns );
 611                                 if(!is_null($varnt)){
 612                                         $linkBatch->addObj($varnt);
 613                                         $titles[]=$varnt;
 614                                 }
 615                         }
 616                 }
 617
 618                 // fetch all variants in single query
 619                 $linkBatch->execute();
 620
 621                 foreach( $titles as $varnt ) {
 622                         if( $varnt->getArticleID() > 0 ) {
 623                                 $nt = $varnt;
 624                                 $link = $varnt->getText();
 625                                 break;
 626                         }
 627                 }
 628         }
 629
 630     /**
 631          * returns language specific hash options
 632          *
 633          * @public
 634          */
 635         function getExtraHashOptions() {
 636                 $variant = $this->getPreferredVariant();
 637                 return '!' . $variant ;
 638         }
 639
 640     /**
 641          * get title text as defined in the body of the article text
 642          *
 643          * @public
 644          */
 645         function getParsedTitle() {
 646                 return $this->mTitleDisplay;
 647         }
 648
 649         /**
 650          * Load default conversion tables
 651          * This method must be implemented in derived class
 652          *
 653          * @private
 654          */
 655         function loadDefaultTables() {
 656                 $name = get_class($this);
 657                 wfDie("Must implement loadDefaultTables() method in class $name");
 658         }
 659
 660         /**
 661          * load conversion tables either from the cache or the disk
 662          * @private
 663          */
 664         function loadTables($fromcache=true) {
 665                 global $wgMemc;
 666                 if( $this->mTablesLoaded )
 667                         return;
 668                 wfProfileIn( __METHOD__ );
 669                 $this->mTablesLoaded = true;
 670                 $this->mTables = false;
 671                 if($fromcache) {
 672                         wfProfileIn( __METHOD__.'-cache' );
 673                         $this->mTables = $wgMemc->get( $this->mCacheKey );
 674                         wfProfileOut( __METHOD__.'-cache' );
 675                 }
 676                 if ( !$this->mTables || !isset( $this->mTables[self::CACHE_VERSION_KEY] ) ) {
 677                         wfProfileIn( __METHOD__.'-recache' );
 678                         // not in cache, or we need a fresh reload.
 679                         // we will first load the default tables
 680                         // then update them using things in MediaWiki:Zhconversiontable/*
 681                         $this->loadDefaultTables();
 682                         foreach($this->mVariants as $var) {
 683                                 $cached = $this->parseCachedTable($var);
 684                                 $this->mTables[$var]->mergeArray($cached);
 685                         }
 686
 687                         $this->postLoadTables();
 688                         $this->mTables[self::CACHE_VERSION_KEY] = true;
 689
 690                         $wgMemc->set($this->mCacheKey, $this->mTables, 43200);
 691                         wfProfileOut( __METHOD__.'-recache' );
 692                 }
 693                 wfProfileOut( __METHOD__ );
 694         }
 695
 696     /**
 697          * Hook for post processig after conversion tables are loaded
 698          *
 699          */
 700         function postLoadTables() {}
 701
 702     /**
 703          * Reload the conversion tables
 704          *
 705          * @private
 706          */
 707         function reloadTables() {
 708                 if($this->mTables)
 709                         unset($this->mTables);
 710                 $this->mTablesLoaded = false;
 711                 $this->loadTables(false);
 712         }
 713
 714
 715         /**
 716          * parse the conversion table stored in the cache
 717          *
 718          * the tables should be in blocks of the following form:
 719          *              -{
 720          *                      word => word ;
 721          *                      word => word ;
 722          *                      ...
 723          *              }-
 724          *
 725          *      to make the tables more manageable, subpages are allowed
 726          *      and will be parsed recursively if $recursive=true
 727          *
 728          */
 729         function parseCachedTable($code, $subpage='', $recursive=true) {
 730                 global $wgMessageCache;
 731                 static $parsed = array();
 732
 733                 if(!is_object($wgMessageCache))
 734                         return array();
 735
 736                 $key = 'Conversiontable/'.$code;
 737                 if($subpage)
 738                         $key .= '/' . $subpage;
 739
 740                 if(array_key_exists($key, $parsed))
 741                         return array();
 742
 743                 if ( strpos( $code, '/' ) === false ) {
 744                         $txt = $wgMessageCache->get( 'Conversiontable', true, $code );
 745                 } else {
 746                         $title = Title::makeTitleSafe( NS_MEDIAWIKI, "Conversiontable/$code" );
 747                         if ( $title && $title->exists() ) {
 748                                 $article = new Article( $title );
 749                                 $txt = $article->getContents();
 750                         } else {
 751                                 $txt = '';
 752                         }
 753                 }
 754
 755                 // get all subpage links of the form
 756                 // [[MediaWiki:conversiontable/zh-xx/...|...]]
 757                 $linkhead = $this->mLangObj->getNsText(NS_MEDIAWIKI) . ':Conversiontable';
 758                 $subs = explode('[[', $txt);
 759                 $sublinks = array();
 760                 foreach( $subs as $sub ) {
 761                         $link = explode(']]', $sub, 2);
 762                         if(count($link) != 2)
 763                                 continue;
 764                         $b = explode('|', $link[0]);
 765                         $b = explode('/', trim($b[0]), 3);
 766                         if(count($b)==3)
 767                                 $sublink = $b[2];
 768                         else
 769                                 $sublink = '';
 770
 771                         if($b[0] == $linkhead && $b[1] == $code) {
 772                                 $sublinks[] = $sublink;
 773                         }
 774                 }
 775
 776
 777                 // parse the mappings in this page
 778                 $blocks = explode($this->mMarkup['begin'], $txt);
 779                 array_shift($blocks);
 780                 $ret = array();
 781                 foreach($blocks as $block) {
 782                         $mappings = explode($this->mMarkup['end'], $block, 2);
 783                         $stripped = str_replace(array("'", '"', '*','#'), '', $mappings[0]);
 784                         $table = explode( ';', $stripped );
 785                         foreach( $table as $t ) {
 786                                 $m = explode( '=>', $t );
 787                                 if( count( $m ) != 2)
 788                                         continue;
 789                                 // trim any trailling comments starting with '//'
 790                                 $tt = explode('//', $m[1], 2);
 791                                 $ret[trim($m[0])] = trim($tt[0]);
 792                         }
 793                 }
 794                 $parsed[$key] = true;
 795
 796
 797                 // recursively parse the subpages
 798                 if($recursive) {
 799                         foreach($sublinks as $link) {
 800                                 $s = $this->parseCachedTable($code, $link, $recursive);
 801                                 $ret = array_merge($ret, $s);
 802                         }
 803                 }
 804
 805                 if ($this->mUcfirst) {
 806                         foreach ($ret as $k => $v) {
 807                                 $ret[Language::ucfirst($k)] = Language::ucfirst($v);
 808                         }
 809                 }
 810                 return $ret;
 811         }
 812
 813         /**
 814          * Enclose a string with the "no conversion" tag. This is used by
 815          * various functions in the Parser
 816          *
 817          * @param string $text text to be tagged for no conversion
 818          * @return string the tagged text
 819          * @public
 820          */
 821         function markNoConversion($text, $noParse=false) {
 822                 # don't mark if already marked
 823                 if(strpos($text, $this->mMarkup['begin']) ||
 824                    strpos($text, $this->mMarkup['end']))
 825                         return $text;
 826
 827                 $ret = $this->mMarkup['begin'] .'R|'. $text . $this->mMarkup['end'];
 828                 return $ret;
 829         }
 830
 831         /**
 832          * convert the sorting key for category links. this should make different
 833          * keys that are variants of each other map to the same key
 834          */
 835         function convertCategoryKey( $key ) {
 836                 return $key;
 837         }
 838         /**
 839          * hook to refresh the cache of conversion tables when
 840          * MediaWiki:conversiontable* is updated
 841          * @private
 842          */
 843         function OnArticleSaveComplete($article, $user, $text, $summary, $isminor, $iswatch, $section, $flags, $revision) {
 844                 $titleobj = $article->getTitle();
 845                 if($titleobj->getNamespace() == NS_MEDIAWIKI) {
 846                         $title = $titleobj->getDBkey();
 847                         $t = explode('/', $title, 3);
 848                         $c = count($t);
 849                         if( $c > 1 && $t[0] == 'Conversiontable' ) {
 850                                 if(in_array($t[1], $this->mVariants)) {
 851                                         $this->reloadTables();
 852                                 }
 853                         }
 854                 }
 855                 return true;
 856         }
 857
 858         /**
 859          * Armour rendered math against conversion
 860          * Wrap math into rawoutput -{R| math }- syntax
 861          * @public
 862          */
 863         function armourMath($text){
 864                 // we need to convert '-{' and '}-' to '-&#123;' and '&#125;-'
 865                 // to avoid a unwanted '}-' appeared after the math-image.
 866                 $text = strtr( $text, array('-{' => '-&#123;', '}-' => '&#125;-') );
 867                 $ret = $this->mMarkup['begin'] . 'R|' . $text . $this->mMarkup['end'];
 868                 return $ret;
 869         }
 870 }
 871
 872 /**
 873  * parser for rules of language conversion , parse rules in -{ }- tag
 874  * @ingroup Language
 875  * @author fdcn <fdcn64@gmail.com>, PhiLiP <philip.npc@gmail.com>
 876  */
 877 class ConverterRule {
 878         var $mText; // original text in -{text}-
 879         var $mConverter; // LanguageConverter object
 880         var $mManualCodeError='<strong class="error">code error!</strong>';
 881         var $mRuleDisplay = '',$mRuleTitle=false;
 882         var $mRules = '';// string : the text of the rules
 883         var $mRulesAction = 'none';
 884         var $mFlags = array();
 885         var $mConvTable = array();
 886         var $mBidtable = array();// array of the translation in each variant
 887         var $mUnidtable = array();// array of the translation in each variant
 888
 889         /**
 890          * Constructor
 891          *
 892          * @param string $text the text between -{ and }-
 893          * @param object $converter a  LanguageConverter object
 894          * @access public
 895          */
 896         function __construct($text,$converter){
 897                 $this->mText = $text;
 898                 $this->mConverter=$converter;
 899                 foreach($converter->mVariants as $v){
 900                         $this->mConvTable[$v]=array();
 901                 }
 902         }
 903
 904         /**
 905          * check if variants array in convert array
 906          *
 907          * @param string $variant Variant language code
 908          * @return string Translated text
 909          * @public
 910          */
 911         function getTextInBidtable($variants){
 912                 if(is_string($variants)){ $variants=array($variants); }
 913                 if(!is_array($variants)) return false;
 914                 foreach ($variants as $variant){
 915                         if(array_key_exists($variant, $this->mBidtable)){
 916                                 return $this->mBidtable[$variant];
 917                         }
 918                 }
 919                 return false;
 920         }
 921
 922         /**
 923          * Parse flags with syntax -{FLAG| ... }-
 924          * @private
 925          */
 926         function parseFlags(){
 927                 $text = $this->mText;
 928                 if(strlen($text) < 2 ) {
 929                         $this->mFlags = array( 'R' );
 930                         $this->mRules = $text;
 931                         return;
 932                 }
 933
 934                 $flags = array();
 935                 $markup = $this->mConverter->mMarkup;
 936                 $validFlags = $this->mConverter->mFlags;
 937                 $variants = $this->mConverter->mVariants;
 938
 939                 $tt = explode($markup['flagsep'], $text, 2);
 940                 if(count($tt) == 2) {
 941                         $f = explode($markup['varsep'], $tt[0]);
 942                         foreach($f as $ff) {
 943                                 $ff = trim($ff);
 944                                 if(array_key_exists($ff, $validFlags) &&
 945                                                         !in_array($validFlags[$ff], $flags))
 946                                         $flags[] = $validFlags[$ff];
 947                         }
 948                         $rules = $tt[1];
 949                 } else {
 950                         $rules = $text;
 951                 }
 952
 953                 //check flags
 954                 if( in_array('R',$flags) ){
 955                         $flags = array('R');// remove other flags
 956                 } elseif ( in_array('N',$flags) ){
 957                         $flags = array('N');// remove other flags
 958                 } elseif ( in_array('-',$flags) ){
 959                         $flags = array('-');// remove other flags
 960                 } elseif (count($flags)==1 && $flags[0]=='T'){
 961                         $flags[]='H';
 962                 } elseif ( in_array('H',$flags) ){
 963                         // replace A flag, and remove other flags except T
 964                         $temp=array('+','H');
 965                         if(in_array('T',$flags)) $temp[] = 'T';
 966                         if(in_array('D',$flags)) $temp[] = 'D';
 967                         $flags = $temp;
 968                 } else {
 969                         if ( in_array('A',$flags) ) {
 970                                 $flags[]='+';
 971                                 $flags[]='S';
 972                         }
 973                         if ( in_array('D',$flags) )
 974                                 $flags=array_diff($flags,array('S'));
 975                         $flags_temp = array();
 976                         foreach ($variants as $variant) {
 977                                 // try to find flags like "zh-hans", "zh-hant"
 978                                 // allow syntaxes like "-{zh-hans;zh-hant|XXXX}-"
 979                                 if ( in_array($variant, $flags) )
 980                                         $flags_temp[] = $variant;
 981                         }
 982                         if ( count($flags_temp) !== 0 )
 983                                 $flags = $flags_temp;
 984                 }
 985                 if ( count($flags) == 0 )
 986                         $flags = array('S');
 987                 $this->mRules=$rules;
 988                 $this->mFlags=$flags;
 989         }
 990
 991         /**
 992          * generate conversion table
 993          * @private
 994          */
 995         function parseRules() {
 996                 $rules = $this->mRules;
 997                 $flags = $this->mFlags;
 998                 $bidtable = array();
 999                 $unidtable = array();
1000                 $markup = $this->mConverter->mMarkup;
1001                 $variants = $this->mConverter->mVariants;
1002
1003                 // varsep_pattern for preg_split:
1004                 // text should be splited by ";" only if a valid variant
1005                 // name exist after the markup, for example:
1006                 //  -{zh-hans:<span style="font-size:120%;">xxx</span>;zh-hant:<span style="font-size:120%;">yyy</span>;}-
1007                 // we should split it as:
1008                 //  array(
1009                 //        [0] => 'zh-hans:<span style="font-size:120%;">xxx</span>'
1010                 //        [1] => 'zh-hant:<span style="font-size:120%;">yyy</span>'
1011                 //        [2] => ''
1012                 //       )
1013                 $varsep_pattern = '/' . $markup['varsep'] . '\s*' . '(?=';
1014                 foreach( $variants as $variant ) {
1015                         $varsep_pattern .= $variant . '\s*' . $markup['codesep'] . '|'; // zh-hans:xxx;zh-hant:yyy
1016                         $varsep_pattern .= '[^;]*?' . $markup['unidsep'] . '\s*' . $variant
1017                                                         . '\s*' . $markup['codesep'] . '|'; // xxx=>zh-hans:yyy; xxx=>zh-hant:zzz
1018                 }
1019                 $varsep_pattern .= '\s*$)/';
1020
1021                 $choice = preg_split($varsep_pattern, $rules);
1022
1023                 foreach( $choice as $c ) {
1024                         $v  = explode($markup['codesep'], $c, 2);
1025                         if( count($v) != 2 )
1026                                 continue;// syntax error, skip
1027                         $to = trim($v[1]);
1028                         $v  = trim($v[0]);
1029                         $u  = explode($markup['unidsep'], $v);
1030                         // if $to is empty, strtr() could return a wrong result
1031                         if( count($u) == 1 && $to && in_array( $v, $variants ) ) {
1032                                 $bidtable[$v] = $to;
1033                         } else if(count($u) == 2){
1034                                 $from = trim($u[0]);
1035                                 $v    = trim($u[1]);
1036                                 if( array_key_exists( $v, $unidtable ) && !is_array( $unidtable[$v] )
1037                                         && $to && in_array( $v, $variants ) )
1038                                         $unidtable[$v] = array( $from=>$to );
1039                                 elseif ( $to && in_array( $v, $variants ) )
1040                                         $unidtable[$v][$from] = $to;
1041                         }
1042                         // syntax error, pass
1043                         if ( !array_key_exists( $v, $this->mConverter->mVariantNames ) ){
1044                                 $bidtable = array();
1045                                 $unidtable = array();
1046                                 break;
1047                         }
1048                 }
1049                 $this->mBidtable = $bidtable;
1050                 $this->mUnidtable = $unidtable;
1051         }
1052
1053         /**
1054          * @private
1055          */
1056         function getRulesDesc(){
1057                 $codesep = $this->mConverter->mDescCodeSep;
1058                 $varsep = $this->mConverter->mDescVarSep;
1059                 $text='';
1060                 foreach($this->mBidtable as $k => $v)
1061                         $text .= $this->mConverter->mVariantNames[$k]."$codesep$v$varsep";
1062                 foreach($this->mUnidtable as $k => $a)
1063                         foreach($a as $from=>$to)
1064                                 $text.=$from.'⇒'.$this->mConverter->mVariantNames[$k]."$codesep$to$varsep";
1065                 return $text;
1066         }
1067
1068         /**
1069          *  Parse rules conversion
1070          * @private
1071          */
1072         function getRuleConvertedStr($variant,$doConvert){
1073                 $bidtable = $this->mBidtable;
1074                 $unidtable = $this->mUnidtable;
1075
1076                 if( count($bidtable) + count($unidtable) == 0 ){
1077                         return $this->mRules;
1078                 } elseif ($doConvert){// the text converted
1079                         // display current variant in bidirectional array
1080                         $disp = $this->getTextInBidtable($variant);
1081                         // or display current variant in fallbacks
1082                         if(!$disp)
1083                                 $disp = $this->getTextInBidtable(
1084                                                 $this->mConverter->getVariantFallbacks($variant));
1085                         // or display current variant in unidirectional array
1086                         if(!$disp && array_key_exists($variant,$unidtable)){
1087                                 $disp = array_values($unidtable[$variant]);
1088                                 $disp = $disp[0];
1089                         }
1090                         // or display frist text under disable manual convert
1091                         if(!$disp && $this->mConverter->mManualLevel[$variant]=='disable') {
1092                                 if(count($bidtable)>0){
1093                                         $disp = array_values($bidtable);
1094                                         $disp = $disp[0];
1095                                 } else {
1096                                         $disp = array_values($unidtable);
1097                                         $disp = array_values($disp[0]);
1098                                         $disp = $disp[0];
1099                                 }
1100                         }
1101                         return $disp;
1102                 } else {// no convert
1103                         return $this->mRules;
1104                 }
1105         }
1106
1107         /**
1108          * generate conversion table for all text
1109          * @private
1110          */
1111         function generateConvTable(){
1112                 $flags = $this->mFlags;
1113                 $bidtable = $this->mBidtable;
1114                 $unidtable = $this->mUnidtable;
1115                 $manLevel = $this->mConverter->mManualLevel;
1116
1117                 $vmarked=array();
1118                 foreach($this->mConverter->mVariants as $v) {
1119                         /* for bidirectional array
1120                                 fill in the missing variants, if any,
1121                                 with fallbacks */
1122                         if(!array_key_exists($v, $bidtable)) {
1123                                 $variantFallbacks = $this->mConverter->getVariantFallbacks($v);
1124                                 $vf = $this->getTextInBidtable($variantFallbacks);
1125                                 if($vf) $bidtable[$v] = $vf;
1126                         }
1127
1128                         if(array_key_exists($v,$bidtable)){
1129                                 foreach($vmarked as $vo){
1130                                         // use syntax: -{A|zh:WordZh;zh-tw:WordTw}-
1131                                         // or -{H|zh:WordZh;zh-tw:WordTw}- or -{-|zh:WordZh;zh-tw:WordTw}-
1132                                         // to introduce a custom mapping between
1133                                         // words WordZh and WordTw in the whole text
1134                                         if($manLevel[$v]=='bidirectional'){
1135                                                 $this->mConvTable[$v][$bidtable[$vo]]=$bidtable[$v];
1136                                         }
1137                                         if($manLevel[$vo]=='bidirectional'){
1138                                                 $this->mConvTable[$vo][$bidtable[$v]]=$bidtable[$vo];
1139                                         }
1140                                 }
1141                                 $vmarked[]=$v;
1142                         }
1143                         /*for unidirectional array
1144                                 fill to convert tables */
1145                         $allow_unid = $manLevel[$v]=='bidirectional'
1146                                         || $manLevel[$v]=='unidirectional';
1147                         if($allow_unid && array_key_exists($v,$unidtable)){
1148                                 $ct=$this->mConvTable[$v];
1149                                 $this->mConvTable[$v] = array_merge($ct,$unidtable[$v]);
1150                         }
1151                 }
1152         }
1153
1154         /**
1155          * Parse rules and flags
1156          * @public
1157          */
1158         function parse($variant){
1159                 if(!$variant)
1160                         $variant = $this->mConverter->getPreferredVariant();
1161
1162                 $variants = $this->mConverter->mVariants;
1163                 $this->parseFlags();
1164                 $flags = $this->mFlags;
1165
1166                 // convert to specified variant
1167                 // syntax: -{zh-hans;zh-hant[;...]|<text to convert>}-
1168                 if( count( array_diff( $flags, $variants ) ) == 0 and count( $flags ) != 0 ) {
1169                         if ( in_array( $variant, $flags ) ) // check if current variant in flags
1170                                 // then convert <text to convert> to current language
1171                                 $this->mRules = $this->mConverter->autoConvert( $this->mRules, $variant );
1172                         else { // if current variant no in flags,
1173                                    // then we check its fallback variants.
1174                                 $variantFallbacks = $this->mConverter->getVariantFallbacks($variant);
1175                                 foreach ( $variantFallbacks as $variantFallback ) {
1176                                         // if current variant's fallback exist in flags
1177                                         if ( in_array( $variantFallback, $flags ) ) {
1178                                                 // then convert <text to convert> to fallback language
1179                                                 $this->mRules = $this->mConverter->autoConvert( $this->mRules, $variantFallback );
1180                                                 break;
1181                                         }
1182                                 }
1183                         }
1184                         $this->mFlags = $flags = array('R');
1185                 }
1186
1187                 if( !in_array( 'R', $flags ) || !in_array( 'N', $flags ) ) {
1188                         // decode => HTML entities modified by Sanitizer::removeHTMLtags
1189                         $this->mRules = str_replace('=&gt;','=>',$this->mRules);
1190
1191                         $this->parseRules();
1192                 }
1193                 $rules = $this->mRules;
1194
1195                 if( count( $this->mBidtable ) == 0 && count( $this->mUnidtable ) == 0 ){
1196                         if(in_array('+',$flags) || in_array('-',$flags))
1197                                 // fill all variants if text in -{A/H/-|text} without rules
1198                                 foreach($this->mConverter->mVariants as $v)
1199                                         $this->mBidtable[$v] = $rules;
1200                         elseif (!in_array('N',$flags) && !in_array('T',$flags) )
1201                                 $this->mFlags = $flags = array('R');
1202                 }
1203
1204                 if( in_array('R',$flags) ) {
1205                         // if we don't do content convert, still strip the -{}- tags
1206                         $this->mRuleDisplay = $rules;
1207                 } elseif ( in_array('N',$flags) ){
1208                         // proces N flag: output current variant name
1209                         $this->mRuleDisplay = $this->mConverter->mVariantNames[trim($rules)];
1210                 } elseif ( in_array('D',$flags) ){
1211                         // proces D flag: output rules description
1212                         $this->mRuleDisplay = $this->getRulesDesc();
1213                 } elseif ( in_array('H',$flags) || in_array('-',$flags) ) {
1214                         // proces H,- flag or T only: output nothing
1215                         $this->mRuleDisplay = '';
1216                 } elseif ( in_array('S',$flags) ){
1217                         $this->mRuleDisplay = $this->getRuleConvertedStr($variant,
1218                                                         $this->mConverter->mDoContentConvert);
1219                 } else {
1220                         $this->mRuleDisplay= $this->mManualCodeError;
1221                 }
1222                 // proces T flag
1223                 if ( in_array('T',$flags) ) {
1224                         $this->mRuleTitle = $this->getRuleConvertedStr($variant,
1225                                                         $this->mConverter->mDoTitleConvert);
1226                 }
1227
1228                 if (in_array('-', $flags))
1229                         $this->mRulesAction='remove';
1230                 if (in_array('+', $flags))
1231                         $this->mRulesAction='add';
1232
1233                 $this->generateConvTable();
1234         }
1235
1236         /**
1237          * @public
1238          */
1239         function hasRules(){
1240                 // TODO:
1241         }
1242
1243         /**
1244          * get display text on markup -{...}-
1245          * @public
1246          */
1247         function getDisplay(){
1248                 return $this->mRuleDisplay;
1249         }
1250         /**
1251          * get converted title
1252          * @public
1253          */
1254         function getTitle(){
1255                 return $this->mRuleTitle;
1256         }
1257
1258         /**
1259          * return how deal with conversion rules
1260          * @public
1261          */
1262         function getRulesAction(){
1263                 return $this->mRulesAction;
1264         }
1265
1266         /**
1267          * get conversion table ( bidirectional and unidirectional conversion table )
1268          * @public
1269          */
1270         function getConvTable(){
1271                 return $this->mConvTable;
1272         }
1273
1274         /**
1275          * get conversion rules string
1276          * @public
1277          */
1278         function getRules(){
1279                 return $this->mRules;
1280         }
1281
1282         /**
1283          * get conversion flags
1284          * @public
1285          */
1286         function getFlags(){
1287                 return $this->mFlags;
1288         }
1289 }