includes/ParserXML.php

   1 <?php
   2 require_once ( "Parser.php" ) ;
   3
   4 /**
   5  * This should one day become the XML->(X)HTML parser
   6  * Based on work by Jan Hidders and Magnus Manske
   7  * To use, set
   8  *    $wgUseXMLparser = true ;
   9  *    $wgEnableParserCache = false ;
  10  *    $wgWiki2xml to the path and executable of the command line version (cli)
  11  * in LocalSettings.php
  12  * @package MediaWiki
  13  * @subpackage Experimental
  14  */
  15
  16 /**
  17  * the base class for an element
  18  */
  19 class element {
  20     var $name = '';
  21     var $attrs = array();
  22     var $children = array();
  23
  24     /**
  25     * This finds the ATTRS element and returns the ATTR sub-children as a single string
  26     */
  27     function getSourceAttrs ()
  28         {
  29         $ret = "" ;
  30         foreach ($this->children as $child)
  31             {
  32             if ( !is_string($child) AND $child->name == "ATTRS" )
  33                 {
  34                 $ret = $child->makeXHTML ( $parser );
  35                 }
  36             }
  37         return $ret ;
  38         }
  39
  40     /**
  41     * This collects the ATTR thingies for getSourceAttrs()
  42     */
  43     function getTheseAttrs ()
  44         {
  45         $ret = array() ;
  46         foreach ($this->children as $child)
  47             {
  48             if ( !is_string($child) AND $child->name == "ATTR" )
  49                 {
  50                 $ret[] = $child->attrs["NAME"] . "='" . $child->children[0] . "'" ;
  51                 }
  52             }
  53         return implode ( " " , $ret ) ;
  54         }
  55
  56     function fixLinkTails ( &$parser , $key )
  57         {
  58         $k2 = $key + 1 ;
  59         if ( !isset ( $this->children[$k2] ) ) return ;
  60         if ( !is_string ( $this->children[$k2]) ) return ;
  61         if ( is_string ( $this->children[$key]) ) return ;
  62         if ( $this->children[$key]->name != "LINK" ) return ;
  63
  64         $n = $this->children[$k2] ;
  65         $s = "" ;
  66         while ( $n != "" AND
  67                 ( ( $n[0] >= 'a' AND $n[0] <= 'z' ) OR
  68                   $n[0] == 'ä' OR $n[0] == 'ö' OR
  69                   $n[0] == 'ü' OR $n[0] == 'ß' ) )
  70               {
  71               $s .= $n[0] ;
  72               $n = substr ( $n , 1 ) ;
  73               }
  74         $this->children[$k2] = $n ;
  75
  76         if ( count ( $this->children[$key]->children ) > 1 )
  77            {
  78            $kl = array_keys ( $this->children[$key]->children ) ;
  79            $kl = array_pop ( $kl ) ;
  80            $this->children[$key]->children[$kl]->children[] = $s ;
  81            }
  82         else
  83            {
  84            $e = new element ;
  85            $e->name = "LINKOPTION" ;
  86            $t = $this->children[$key]->sub_makeXHTML ( $parser ) ;
  87            $e->children[] = trim ( $t ) . $s ;
  88            $this->children[$key]->children[] = $e ;
  89            }
  90         }
  91
  92     /**
  93     * This function generates the XHTML for the entire subtree
  94     */
  95     function sub_makeXHTML ( &$parser , $tag = "" , $attr = "" )
  96         {
  97         $ret = "" ;
  98
  99         $attr2 = $this->getSourceAttrs () ;
 100         if ( $attr != "" AND $attr2 != "" ) $attr .= " " ;
 101         $attr .= $attr2 ;
 102
 103         if ( $tag != "" )
 104                 {
 105                 $ret .= "<" . $tag ;
 106                 if ( $attr != "" ) $ret .= " " . $attr ;
 107                 $ret .= ">" ;
 108                 }
 109
 110 #       foreach ( array_keys ( $this->children ) AS $x )
 111 #          $this->fixLinkTails ( $parser , $x ) ;
 112
 113         foreach ($this->children as $key => $child) {
 114             if ( is_string($child) ) {
 115                 $ret .= $child ;
 116             } else if ( $child->name != "ATTRS" ) {
 117                 $ret .= $child->makeXHTML ( $parser );
 118             }
 119            }
 120         if ( $tag != "" )
 121                 $ret .= "</" . $tag . ">\n" ;
 122         return $ret ;
 123         }
 124
 125     /**
 126     * Link functions
 127     */
 128     function createInternalLink ( &$parser , $target , $display_title , $options )
 129         {
 130         global $wgUser ;
 131         $skin = $wgUser->getSkin() ;
 132         $tp = explode ( ":" , $target ) ; # tp = target parts
 133         $title = "" ;     # The plain title
 134         $language = "" ;  # The language/meta/etc. part
 135         $namespace = "" ; # The namespace, if any
 136         $subtarget = "" ; # The '#' thingy
 137
 138
 139         $nt = Title::newFromText ( $target ) ;
 140         $fl = strtoupper ( $this->attrs["FORCEDLINK"] ) == "YES" ;
 141
 142         if ( $fl || count ( $tp ) == 1 ) $title = $target ; # Plain and simple case
 143         else # There's stuff missing here...
 144                 {
 145                 if ( $nt->getNamespace() == NS_IMAGE )
 146                    {
 147                    $options[] = $display_title ;
 148                    return $skin->makeImageLinkObj ( $nt , implode ( "|" , $options ) ) ;
 149                    }
 150                 else $title = $target ; # Default
 151                 }
 152
 153         if ( $language != "" ) # External link within the WikiMedia project
 154                 {
 155                 return "{language link}" ;
 156                 }
 157         else if ( $namespace != "" ) # Link to another namespace, check for image/media stuff
 158                 {
 159                 return "{namespace link}" ;
 160                 }
 161         else
 162                 {
 163                 return $skin->makeLink ( $target , $display_title ) ;
 164                 }
 165         }
 166
 167     function makeInternalLink ( &$parser )
 168         {
 169         $target = "" ;
 170         $option = array () ;
 171       foreach ($this->children as $child) {
 172             if ( is_string($child) ) {
 173                 # This shouldn't be the case!
 174             } else {
 175                 if ( $child->name == "LINKTARGET" )
 176                         $target = trim ( $child->makeXHTML ( $parser ) ) ;
 177                     else
 178                         $option[] = trim ( $child->makeXHTML ( $parser ) ) ;
 179             }
 180            }
 181
 182       if ( count ( $option ) == 0 ) $option[] = $target ; # Create dummy display title
 183       $display_title = array_pop ( $option ) ;
 184       return $this->createInternalLink ( $parser , $target , $display_title , $option ) ;
 185         }
 186
 187     function getTemplateXHTML ( $title , $parts , &$parser ) {
 188         global $wgLang , $wgUser ;
 189         $skin = $wgUser->getSkin() ;
 190         $ot = $title ; # Original title
 191         if ( count ( explode ( ":" , $title ) ) == 1 )
 192            $title = $wgLang->getNsText ( NS_TEMPLATE ) . ":" . $title ;
 193         $nt = Title::newFromText ( $title ) ;
 194         $id = $nt->getArticleID() ;
 195         if ( $id == 0 ) { # No/non-existing page
 196            return $skin->makeBrokenLink ( $title , $ot ) ;
 197            }
 198
 199         $a = 0 ;
 200         $tv = array () ; # Template variables
 201         foreach ( $parts AS $part ) {
 202            $a++ ;
 203            $x = explode ( "=" , $part , 2 ) ;
 204            if ( count ( $x ) == 1 ) $key = "{$a}" ;
 205            else $key = $x[0] ;
 206            $value = array_pop ( $x ) ;
 207            $tv[$key] = $value ;
 208            }
 209         $art = new Article ( $nt ) ;
 210         $text = $art->getContent ( false ) ;
 211         $parser->plain_parse ( $text , true , $tv ) ;
 212
 213         return $text ;
 214     }
 215
 216     /**
 217     * This function actually converts wikiXML into XHTML tags
 218     */
 219     function makeXHTML ( &$parser )
 220         {
 221         $ret = "" ;
 222         $n = $this->name ; # Shortcut
 223
 224         if ( $n == "EXTENSION" ) # Fix allowed HTML
 225            {
 226            $old_n = $n ;
 227            $ext = strtoupper ( $this->attrs["NAME"] ) ;
 228            if ( $ext == "B" || $ext == "STRONG" ) $n = "BOLD" ;
 229            else if ( $ext == "I" || $ext == "EM" ) $n = "ITALICS" ;
 230            else if ( $ext == "U" ) $n = "UNDERLINED" ; # Hey, virtual wiki tag! ;-)
 231            else if ( $ext == "S" ) $n = "STRIKE" ;
 232            else if ( $ext == "P" ) $n = "PARAGRAPH" ;
 233            else if ( $ext == "TABLE" ) $n = "TABLE" ;
 234            else if ( $ext == "TR" ) $n = "TABLEROW" ;
 235            else if ( $ext == "TD" ) $n = "TABLECELL" ;
 236            else if ( $ext == "TH" ) $n = "TABLEHEAD" ;
 237            else if ( $ext == "CAPTION" ) $n = "CAPTION" ;
 238            else if ( $ext == "NOWIKI" ) $n = "NOWIKI" ;
 239            if ( $n != $old_n ) unset ( $this->attrs["NAME"] ) ; # Cleanup
 240            else if ( $parser->nowiki > 0 ) $n = "" ; # No "real" wiki tags allowed in nowiki section
 241            }
 242
 243         if ( $n == "ARTICLE" )
 244                 $ret .= $this->sub_makeXHTML ( $parser ) ;
 245         else if ( $n == "HEADING" )
 246                 $ret .= $this->sub_makeXHTML ( $parser , "h" . $this->attrs["LEVEL"] ) ;
 247         else if ( $n == "PARAGRAPH" )
 248                 $ret .= $this->sub_makeXHTML ( $parser , "p" ) ;
 249         else if ( $n == "BOLD" )
 250                 $ret .= $this->sub_makeXHTML ( $parser , "strong" ) ;
 251         else if ( $n == "ITALICS" )
 252                 $ret .= $this->sub_makeXHTML ( $parser , "em" ) ;
 253
 254         # These don't exist as wiki markup
 255         else if ( $n == "UNDERLINED" )
 256                 $ret .= $this->sub_makeXHTML ( $parser , "u" ) ;
 257         else if ( $n == "STRIKE" )
 258                 $ret .= $this->sub_makeXHTML ( $parser , "strike" ) ;
 259
 260         # HTML comment
 261         else if ( $n == "COMMENT" )
 262                 $ret .= "" ; # Comments are parsed out
 263
 264         # Links
 265         else if ( $n == "LINK" )
 266                 $ret .= $this->makeInternalLink ( $parser ) ;
 267         else if ( $n == "LINKTARGET" )
 268                 $ret .= $this->sub_makeXHTML ( $parser ) ;
 269         else if ( $n == "LINKOPTION" )
 270                 $ret .= $this->sub_makeXHTML ( $parser ) ;
 271
 272         else if ( $n == "TEMPLATE" )
 273                 {
 274                 $parts = $this->sub_makeXHTML ( $parser ) ;
 275                 $parts = explode ( "|" , $parts ) ;
 276                 $title = array_shift ( $parts ) ;
 277                 $ret .= $this->getTemplateXHTML ( $title , $parts , &$parser ) ;
 278                 }
 279         else if ( $n == "TEMPLATEVAR" )
 280                 {
 281                 $x = $this->sub_makeXHTML ( $parser ) ;
 282                 if ( isset ( $parser->mCurrentTemplateOptions["{$x}"] ) )
 283                    $ret .= $parser->mCurrentTemplateOptions["{$x}"] ;
 284                 }
 285
 286         else if ( $n == "IGNORE" ) # Internal use, not generated by wiki2xml parser
 287                 $ret .= $this->sub_makeXHTML ( $parser ) ;
 288
 289         else if ( $n == "NOWIKI" )
 290                 {
 291                 $parser->nowiki++ ;
 292                 $ret .= $this->sub_makeXHTML ( $parser , "" ) ;
 293                 $parser->nowiki-- ;
 294                 }
 295
 296         # Unknown HTML extension
 297         else if ( $n == "EXTENSION" ) # This is currently a dummy!!!
 298                 {
 299                 $ext = $this->attrs["NAME"] ;
 300
 301                 $ret .= "&lt;" . $ext . "&gt;" ;
 302                 $ret .= $this->sub_makeXHTML ( $parser ) ;
 303                 $ret .= "&lt;/" . $ext . "&gt; " ;
 304                 }
 305
 306         # Table stuff
 307         else if ( $n == "TABLE" )
 308                 {
 309                 $ret .= $this->sub_makeXHTML ( $parser , "table" ) ;
 310                 }
 311         else if ( $n == "TABLEROW" )
 312                 {
 313                 $ret .= $this->sub_makeXHTML ( $parser , "tr" ) ;
 314                 }
 315         else if ( $n == "TABLECELL" )
 316                 {
 317                 $ret .= $this->sub_makeXHTML ( $parser , "td" ) ;
 318                 }
 319         else if ( $n == "TABLEHEAD" )
 320                 {
 321                 $ret .= $this->sub_makeXHTML ( $parser , "th" ) ;
 322                 }
 323         else if ( $n == "CAPTION" )
 324                 {
 325                 $ret .= $this->sub_makeXHTML ( $parser , "caption" ) ;
 326                 }
 327
 328         else if ( $n == "ATTRS" ) # SPECIAL CASE : returning attributes
 329                 {
 330                 return $this->getTheseAttrs () ;
 331                 }
 332
 333         # Lists
 334         else if ( $n == "LISTITEM" )
 335                 {
 336                 if ( $parser->mListType == "dl" ) $ret .= $this->sub_makeXHTML ( $parser , "dd" ) ;
 337                 else $ret .= $this->sub_makeXHTML ( $parser , "li" ) ;
 338                 }
 339         else if ( $n == "LIST" )
 340                 {
 341                 $type = "ol" ; # Default
 342                 if ( $this->attrs["TYPE"] == "bullet" ) $type = "ul" ;
 343                 else if ( $this->attrs["TYPE"] == "indent" ) $type = "dl" ;
 344                 $oldtype = $parser->mListType ;
 345                 $parser->mListType = $type ;
 346                 $ret .= $this->sub_makeXHTML ( $parser , $type ) ;
 347                 $parser->mListType = $oldtype ;
 348                 }
 349
 350         # Something else entirely
 351         else
 352                 {
 353                 $ret .= "&lt;" . $n . "&gt;" ;
 354                 $ret .= $this->sub_makeXHTML ( $parser ) ;
 355                 $ret .= "&lt;/" . $n . "&gt; " ;
 356                 }
 357
 358         $ret = "\n{$ret}\n" ;
 359         $ret = str_replace ( "\n\n" , "\n" , $ret ) ;
 360         return $ret ;
 361         }
 362
 363     /**
 364     * A function for additional debugging output
 365     */
 366     function myPrint() {
 367         $ret = "<ul>\n";
 368         $ret .= "<li> <b> Name: </b> $this->name </li>\n";
 369         // print attributes
 370         $ret .= '<li> <b> Attributes: </b>';
 371         foreach ($this->attrs as $name => $value) {
 372             $ret .= "$name => $value; " ;
 373         }
 374         $ret .= " </li>\n";
 375         // print children
 376         foreach ($this->children as $child) {
 377             if ( is_string($child) ) {
 378                 $ret .= "<li> $child </li>\n";
 379             } else {
 380                 $ret .= $child->myPrint();
 381             }
 382         }
 383         $ret .= "</ul>\n";
 384         return $ret;
 385     }
 386 }
 387
 388 $ancStack = array();    // the stack with ancestral elements
 389
 390 // Three global functions needed for parsing, sorry guys
 391 function wgXMLstartElement($parser, $name, $attrs) {
 392     global $ancStack;
 393
 394     $newElem = new element;
 395     $newElem->name = $name;
 396     $newElem->attrs = $attrs;
 397
 398     array_push($ancStack, $newElem);
 399 }
 400
 401 function wgXMLendElement($parser, $name) {
 402     global $ancStack, $rootElem;
 403     // pop element off stack
 404     $elem = array_pop ($ancStack);
 405     if (count ($ancStack) == 0)
 406         $rootElem = $elem;
 407     else
 408         // add it to its parent
 409         array_push ($ancStack[count($ancStack)-1]->children, $elem);
 410 }
 411
 412 function wgXMLcharacterData($parser, $data) {
 413     global $ancStack;
 414     $data = trim ($data); // Don't add blank lines, they're no use...
 415     // add to parent if parent exists
 416     if ( $ancStack && $data != "" ) {
 417         array_push ($ancStack[count($ancStack)-1]->children, $data);
 418     }
 419 }
 420
 421
 422 /**
 423  * Here's the class that generates a nice tree
 424  */
 425 class xml2php {
 426
 427     function &scanFile( $filename ) {
 428         global $ancStack, $rootElem;
 429         $ancStack = array();
 430
 431         $xml_parser = xml_parser_create();
 432         xml_set_element_handler ($xml_parser, 'wgXMLstartElement', 'wgXMLendElement');
 433         xml_set_character_data_handler ($xml_parser, 'wgXMLcharacterData');
 434         if (!($fp = fopen($filename, 'r'))) {
 435             die('could not open XML input');
 436         }
 437         while ($data = fread($fp, 4096)) {
 438             if (!xml_parse($xml_parser, $data, feof($fp))) {
 439                 die(sprintf("XML error: %s at line %d",
 440                     xml_error_string(xml_get_error_code($xml_parser)),
 441                     xml_get_current_line_number($xml_parser)));
 442             }
 443         }
 444         xml_parser_free($xml_parser);
 445
 446         // return the remaining root element we copied in the beginning
 447         return $rootElem;
 448     }
 449
 450     function scanString ( $input ) {
 451         global $ancStack, $rootElem;
 452         $ancStack = array();
 453
 454         $xml_parser = xml_parser_create();
 455         xml_set_element_handler ($xml_parser, 'wgXMLstartElement', 'wgXMLendElement');
 456         xml_set_character_data_handler ($xml_parser, 'wgXMLcharacterData');
 457
 458         if (!xml_parse ($xml_parser, $input, true)) {
 459             die (sprintf ("XML error: %s at line %d",
 460                 xml_error_string(xml_get_error_code($xml_parser)),
 461                 xml_get_current_line_number($xml_parser)));
 462         }
 463         xml_parser_free ($xml_parser);
 464
 465         // return the remaining root element we copied in the beginning
 466         return $rootElem;
 467     }
 468
 469 }
 470
 471 class ParserXML EXTENDS Parser
 472         {
 473         /**#@+
 474          * @access private
 475          */
 476         # Persistent:
 477         var $mTagHooks, $mListType;
 478
 479         # Cleared with clearState():
 480         var $mOutput, $mAutonumber, $mDTopen, $mStripState = array();
 481         var $mVariables, $mIncludeCount, $mArgStack, $mLastSection, $mInPre;
 482
 483         # Temporary:
 484         var $mOptions, $mTitle, $mOutputType,
 485             $mTemplates,        // cache of already loaded templates, avoids
 486                                 // multiple SQL queries for the same string
 487             $mTemplatePath;     // stores an unsorted hash of all the templates already loaded
 488                                 // in this path. Used for loop detection.
 489
 490         var $nowikicount , $mCurrentTemplateOptions ;
 491
 492         /**#@-*/
 493
 494         /**
 495          * Constructor
 496          *
 497          * @access public
 498          */
 499         function ParserXML() {
 500                 $this->mTemplates = array();
 501                 $this->mTemplatePath = array();
 502                 $this->mTagHooks = array();
 503                 $this->clearState();
 504         }
 505
 506         /**
 507          * Clear Parser state
 508          *
 509          * @access private
 510          */
 511         function clearState() {
 512                 $this->mOutput = new ParserOutput;
 513                 $this->mAutonumber = 0;
 514                 $this->mLastSection = "";
 515                 $this->mDTopen = false;
 516                 $this->mVariables = false;
 517                 $this->mIncludeCount = array();
 518                 $this->mStripState = array();
 519                 $this->mArgStack = array();
 520                 $this->mInPre = false;
 521         }
 522
 523         /**
 524         * Turns the wikitext into XML by calling the external parser
 525         *
 526         */
 527         function runXMLparser ( &$text ) {
 528                 global $wgWiki2xml ;
 529
 530                 $tmpfname = tempnam("/tmp", "FOO");
 531                 $handle = fopen($tmpfname, "w");
 532                 fwrite($handle, $text);
 533                 fclose($handle);
 534                 exec ( $wgWiki2xml . " < " . $tmpfname , $a ) ;
 535                 $text = implode ( "\n" , $a ) ;
 536                 unlink($tmpfname);
 537         }
 538
 539         function plain_parse ( &$text , $inline = false , $templateOptions = array () ) {
 540                 $this->runXMLparser ( $text ) ;
 541                 $nowikicount = 0 ;
 542                 $w = new xml2php;
 543                 $result = $w->scanString( $text );
 544
 545                 $oldTemplateOptions = $this->mCurrentTemplateOptions ;
 546                 $this->mCurrentTemplateOptions = $templateOptions ;
 547
 548                 if ( $inline ) { # Inline rendering off for templates
 549                    if ( count ( $result->children ) == 1 )
 550                       $result->children[0]->name = "IGNORE" ;
 551                    }
 552
 553                 if ( 1 ) $text = $result->makeXHTML ( $this ) ; # No debugging info
 554                 else $text = $result->makeXHTML ( $this ) . "<hr>" . $text . "<hr>" . $result->myPrint();
 555                 $this->mCurrentTemplateOptions = $oldTemplateOptions ;
 556         }
 557
 558         function parse( $text, &$title, $options, $linestart = true, $clearState = true ) {
 559                 $this->plain_parse ( $text ) ;
 560                 $this->mOutput->setText ( $text ) ;
 561                 return $this->mOutput;
 562         }
 563
 564         }
 565
 566 ?>