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