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 , $option ) ;
 146         }
 147
 148     /**
 149     * This function actually converts wikiXML into XHTML tags
 150     */
 151     function makeXHTML ( &$parser )
 152         {
 153         $ret = "" ;
 154         $n = $this->name ; # Shortcut
 155
 156         if ( $n == "EXTENSION" ) # Fix allowed HTML
 157            {
 158            $old_n = $n ;
 159            $ext = strtoupper ( $this->attrs["NAME"] ) ;
 160            if ( $ext == "B" || $ext == "STRONG" ) $n = "BOLD" ;
 161            else if ( $ext == "I" || $ext == "EM" ) $n = "ITALICS" ;
 162            else if ( $ext == "U" ) $n = "UNDERLINED" ; # Hey, virtual wiki tag! ;-)
 163            else if ( $ext == "S" ) $n = "STRIKE" ;
 164            else if ( $ext == "P" ) $n = "PARAGRAPH" ;
 165            else if ( $ext == "TABLE" ) $n = "TABLE" ;
 166            else if ( $ext == "TR" ) $n = "TABLEROW" ;
 167            else if ( $ext == "TD" ) $n = "TABLECELL" ;
 168            else if ( $ext == "TH" ) $n = "TABLEHEAD" ;
 169            else if ( $ext == "CAPTION" ) $n = "CAPTION" ;
 170            else if ( $ext == "NOWIKI" ) $n = "NOWIKI" ;
 171            if ( $n != $old_n ) unset ( $this->attrs["NAME"] ) ; # Cleanup
 172            else if ( $parser->nowiki > 0 ) $n = "" ; # No "real" wiki tags allowed
 173            }
 174
 175         if ( $n == "ARTICLE" )
 176                 $ret .= $this->sub_makeXHTML ( $parser ) ;
 177         else if ( $n == "HEADING" )
 178                 $ret .= $this->sub_makeXHTML ( $parser , "h" . $this->attrs["LEVEL"] ) ;
 179         else if ( $n == "PARAGRAPH" )
 180                 $ret .= $this->sub_makeXHTML ( $parser , "p" ) ;
 181         else if ( $n == "BOLD" )
 182                 $ret .= $this->sub_makeXHTML ( $parser , "strong" ) ;
 183         else if ( $n == "ITALICS" )
 184                 $ret .= $this->sub_makeXHTML ( $parser , "em" ) ;
 185
 186         # These don't exist as wiki markup
 187         else if ( $n == "UNDERLINED" )
 188                 $ret .= $this->sub_makeXHTML ( $parser , "u" ) ;
 189         else if ( $n == "STRIKE" )
 190                 $ret .= $this->sub_makeXHTML ( $parser , "strike" ) ;
 191
 192         # Links
 193         else if ( $n == "LINK" )
 194                 $ret .= $this->makeInternalLink ( $parser ) ;
 195         else if ( $n == "LINKTARGET" )
 196                 $ret .= $this->sub_makeXHTML ( $parser ) ;
 197         else if ( $n == "LINKOPTION" )
 198                 $ret .= $this->sub_makeXHTML ( $parser ) ;
 199
 200         else if ( $n == "NOWIKI" )
 201                 {
 202                 $parser->nowiki++ ;
 203                 $ret .= $this->sub_makeXHTML ( $parser , "" ) ;
 204                 $parser->nowiki-- ;
 205                 }
 206
 207         # Unknown HTML extension
 208         else if ( $n == "EXTENSION" ) # This is currently a dummy!!!
 209                 {
 210                 $ext = $this->attrs["NAME"] ;
 211
 212                 $ret .= "&lt;" . $ext . "&gt;" ;
 213                 $ret .= $this->sub_makeXHTML ( $parser ) ;
 214                 $ret .= "&lt;/" . $ext . "&gt; " ;
 215                 }
 216
 217         # Table stuff
 218         else if ( $n == "TABLE" )
 219                 {
 220                 $ret .= $this->sub_makeXHTML ( $parser , "table" ) ;
 221                 }
 222         else if ( $n == "TABLEROW" )
 223                 {
 224                 $ret .= $this->sub_makeXHTML ( $parser , "tr" ) ;
 225                 }
 226         else if ( $n == "TABLECELL" )
 227                 {
 228                 $ret .= $this->sub_makeXHTML ( $parser , "td" ) ;
 229                 }
 230         else if ( $n == "TABLEHEAD" )
 231                 {
 232                 $ret .= $this->sub_makeXHTML ( $parser , "th" ) ;
 233                 }
 234         else if ( $n == "CAPTION" )
 235                 {
 236                 $ret .= $this->sub_makeXHTML ( $parser , "caption" ) ;
 237                 }
 238
 239         else if ( $n == "ATTRS" ) # SPECIAL CASE : returning attributes
 240                 {
 241                 return $this->getTheseAttrs () ;
 242                 }
 243
 244         # Lists
 245         else if ( $n == "LISTITEM" )
 246                 $ret .= $this->sub_makeXHTML ( $parser , "li" ) ;
 247         else if ( $n == "LIST" )
 248                 {
 249                 $type = "ol" ; # Default
 250                 if ( $this->attrs["TYPE"] == "bullet" ) $type = "ul" ;
 251                 $ret .= $this->sub_makeXHTML ( $parser , $type ) ;
 252                 }
 253
 254         # Something else entirely
 255         else
 256                 {
 257                 $ret .= "&lt;" . $n . "&gt;" ;
 258                 $ret .= $this->sub_makeXHTML ( $parser ) ;
 259                 $ret .= "&lt;/" . $n . "&gt; " ;
 260                 }
 261
 262         $ret = "\n{$ret}\n" ;
 263         $ret = str_replace ( "\n\n" , "\n" , $ret ) ;
 264         return $ret ;
 265         }
 266
 267     /**
 268     * A function for additional debugging output
 269     */
 270     function myPrint() {
 271         $ret = "<ul>\n";
 272         $ret .= "<li> <b> Name: </b> $this->name </li>\n";
 273         // print attributes
 274         $ret .= '<li> <b> Attributes: </b>';
 275         foreach ($this->attrs as $name => $value) {
 276             $ret .= "$name => $value; " ;
 277         }
 278         $ret .= " </li>\n";
 279         // print children
 280         foreach ($this->children as $child) {
 281             if ( is_string($child) ) {
 282                 $ret .= "<li> $child </li>\n";
 283             } else {
 284                 $ret .= $child->myPrint();
 285             }
 286         }
 287         $ret .= "</ul>\n";
 288         return $ret;
 289     }
 290 }
 291
 292 $ancStack = array();    // the stack with ancestral elements
 293
 294 // Three global functions needed for parsing, sorry guys
 295 function wgXMLstartElement($parser, $name, $attrs) {
 296     global $ancStack;
 297
 298     $newElem = new element;
 299     $newElem->name = $name;
 300     $newElem->attrs = $attrs;
 301
 302     array_push($ancStack, $newElem);
 303 }
 304
 305 function wgXMLendElement($parser, $name) {
 306     global $ancStack, $rootElem;
 307     // pop element off stack
 308     $elem = array_pop ($ancStack);
 309     if (count ($ancStack) == 0)
 310         $rootElem = $elem;
 311     else
 312         // add it to its parent
 313         array_push ($ancStack[count($ancStack)-1]->children, $elem);
 314 }
 315
 316 function wgXMLcharacterData($parser, $data) {
 317     global $ancStack;
 318     $data = trim ($data); // Don't add blank lines, they're no use...
 319     // add to parent if parent exists
 320     if ( $ancStack && $data != "" ) {
 321         array_push ($ancStack[count($ancStack)-1]->children, $data);
 322     }
 323 }
 324
 325
 326 /**
 327  * Here's the class that generates a nice tree
 328  */
 329 class xml2php {
 330
 331     function &scanFile( $filename ) {
 332         global $ancStack, $rootElem;
 333         $ancStack = array();
 334
 335         $xml_parser = xml_parser_create();
 336         xml_set_element_handler ($xml_parser, 'wgXMLstartElement', 'wgXMLendElement');
 337         xml_set_character_data_handler ($xml_parser, 'wgXMLcharacterData');
 338         if (!($fp = fopen($filename, 'r'))) {
 339             die('could not open XML input');
 340         }
 341         while ($data = fread($fp, 4096)) {
 342             if (!xml_parse($xml_parser, $data, feof($fp))) {
 343                 die(sprintf("XML error: %s at line %d",
 344                     xml_error_string(xml_get_error_code($xml_parser)),
 345                     xml_get_current_line_number($xml_parser)));
 346             }
 347         }
 348         xml_parser_free($xml_parser);
 349
 350         // return the remaining root element we copied in the beginning
 351         return $rootElem;
 352     }
 353
 354     function scanString ( $input ) {
 355         global $ancStack, $rootElem;
 356         $ancStack = array();
 357
 358         $xml_parser = xml_parser_create();
 359         xml_set_element_handler ($xml_parser, 'wgXMLstartElement', 'wgXMLendElement');
 360         xml_set_character_data_handler ($xml_parser, 'wgXMLcharacterData');
 361
 362         if (!xml_parse ($xml_parser, $input, true)) {
 363             die (sprintf ("XML error: %s at line %d",
 364                 xml_error_string(xml_get_error_code($xml_parser)),
 365                 xml_get_current_line_number($xml_parser)));
 366         }
 367         xml_parser_free ($xml_parser);
 368
 369         // return the remaining root element we copied in the beginning
 370         return $rootElem;
 371     }
 372
 373 }
 374
 375 /* Example code:
 376
 377     $w = new xml2php;
 378     $filename = 'sample.xml';
 379     $result = $w->scanFile( $filename );
 380     print $result->myPrint();
 381 */
 382
 383 $dummytext = "<article><heading level='2'> R-type </heading><paragraph><link><linktarget>image:a.jpg</linktarget><linkoption>1</linkoption><linkoption>2</linkoption><linkoption>3</linkoption><linkoption>text</linkoption></link></paragraph><paragraph>The <link><linktarget>video game</linktarget><linkoption>computer game</linkoption></link> <bold>R-type</bold> is <extension name='nowiki'>cool &amp; stuff</extension> because:</paragraph><list type='bullet'><listitem>it's nice</listitem><listitem>it's fast</listitem><listitem>it has:<list type='bullet'><listitem>graphics</listitem><listitem>sound</listitem></list></listitem></list><table><tablerow><tablecell>Version 1     </tablecell><tablecell>not bad</tablecell></tablerow><tablerow><tablecell>Version 2     </tablecell><tablecell>much better </tablecell></tablerow></table><paragraph>This is a || token in the middle of text.</paragraph></article>" ;
 384
 385 class ParserXML EXTENDS Parser
 386         {
 387         /**#@+
 388          * @access private
 389          */
 390         # Persistent:
 391         var $mTagHooks;
 392
 393         # Cleared with clearState():
 394         var $mOutput, $mAutonumber, $mDTopen, $mStripState = array();
 395         var $mVariables, $mIncludeCount, $mArgStack, $mLastSection, $mInPre;
 396
 397         # Temporary:
 398         var $mOptions, $mTitle, $mOutputType,
 399             $mTemplates,        // cache of already loaded templates, avoids
 400                                 // multiple SQL queries for the same string
 401             $mTemplatePath;     // stores an unsorted hash of all the templates already loaded
 402                                 // in this path. Used for loop detection.
 403
 404         var $nowikicount ;
 405
 406         /**#@-*/
 407
 408         /**
 409          * Constructor
 410          *
 411          * @access public
 412          */
 413         function ParserXML() {
 414                 $this->mTemplates = array();
 415                 $this->mTemplatePath = array();
 416                 $this->mTagHooks = array();
 417                 $this->clearState();
 418         }
 419
 420         /**
 421          * Clear Parser state
 422          *
 423          * @access private
 424          */
 425         function clearState() {
 426                 $this->mOutput = new ParserOutput;
 427                 $this->mAutonumber = 0;
 428                 $this->mLastSection = "";
 429                 $this->mDTopen = false;
 430                 $this->mVariables = false;
 431                 $this->mIncludeCount = array();
 432                 $this->mStripState = array();
 433                 $this->mArgStack = array();
 434                 $this->mInPre = false;
 435         }
 436
 437         /**
 438         * Turns the wikitext into XML by calling the external parser
 439         *
 440         */
 441         function runXMLparser ( &$text ) {
 442                 global $wgWiki2xml ;
 443
 444                 $tmpfname = tempnam("/tmp", "FOO");
 445                 $handle = fopen($tmpfname, "w");
 446                 fwrite($handle, $text);
 447                 fclose($handle);
 448                 exec ( $wgWiki2xml . " < " . $tmpfname , $a ) ;
 449                 $text = implode ( "\n" , $a ) ;
 450                 unlink($tmpfname);
 451         }
 452
 453         function parse( $text, &$title, $options, $linestart = true, $clearState = true ) {
 454                 $this->runXMLparser ( $text ) ;
 455                 $nowikicount = 0 ;
 456                 $w = new xml2php;
 457                 $result = $w->scanString( $text );
 458
 459                 if ( 1 ) $text = $result->makeXHTML ( $this ) ; # No debugging info
 460                 else $text = $result->makeXHTML ( $this ) . "<hr>" . $text . "<hr>" . $result->myPrint();
 461
 462                 $this->mOutput->setText ( $text ) ;
 463                 return $this->mOutput;
 464         }
 465
 466         }
 467
 468 ?>