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