includes/ParserXML.php

   1 <?
   2 # This should one day become the XML->(X)HTML parser
   3 # Based on work by Jan Hidders and Magnus Manske
   4
   5 //  the base class for an element
   6 class element {
   7   var $name = '';
   8   var $attrs = array();
   9   var $children = array();
  10
  11   function myPrint() {
  12     echo '<UL>';
  13     echo "<LI> <B> Name: </B> $this->name";
  14     // print attributes
  15     echo '<LI> <B> Attributes: </B>';
  16     foreach ($this->attrs as $name => $value) {
  17       echo "$name => $value; " ;
  18     }
  19     // print children
  20     foreach ($this->children as $child) {
  21        if ( is_string($child) ) {
  22          echo '<LI> '.$child;
  23        } else {
  24          $child->myPrint();
  25        }
  26     }
  27     echo '</UL>';
  28   }
  29
  30 }
  31
  32 $ancStack = array();    // the stack with ancestral elements
  33
  34 // Three global functions needed for parsing, sorry guys
  35 function wgXMLstartElement($parser, $name, $attrs) {
  36    global $ancStack, $rootElem;
  37
  38    $newElem = new element;
  39    $newElem->name = $name;
  40    $newElem->attrs = $attrs;
  41    array_push($ancStack, $newElem);
  42    // add to parent if parent exists
  43    $nrAncs = count($ancStack)-1;
  44    if ( $nrAncs > 0 ) {
  45      array_push($ancStack[$nrAncs-1]->children, &$ancStack[$nrAncs]);
  46    } else {
  47      // make extra copy of root element and alias it with the original
  48      array_push($ancStack, &$ancStack[0]);
  49    }
  50 }
  51
  52 function wgXMLendElement($parser, $name) {
  53   global $ancStack, $rootElem;
  54
  55   // pop element of stack
  56   array_pop($ancStack);
  57 }
  58
  59 function wgXMLcharacterData($parser, $data) {
  60   global $ancStack, $rootElem;
  61
  62   $data = trim ( $data ) ; // Don't add blank lines, they're no use...
  63
  64   // add to parent if parent exists
  65   if ( $ancStack && $data != "" ) {
  66     array_push($ancStack[count($ancStack)-1]->children, $data);
  67   }
  68 }
  69
  70
  71 # Here's the class that generates a nice tree
  72 class xml2php {
  73
  74    function &scanFile( $filename ) {
  75        global $ancStack;
  76        $ancStack = array();
  77
  78        $xml_parser = xml_parser_create();
  79        xml_set_element_handler($xml_parser, 'wgXMLstartElement', 'wgXMLendElement');
  80        xml_set_character_data_handler($xml_parser, 'wgXMLcharacterData');
  81        if (!($fp = fopen($filename, 'r'))) {
  82           die('could not open XML input');
  83        }
  84        while ($data = fread($fp, 4096)) {
  85           if (!xml_parse($xml_parser, $data, feof($fp))) {
  86               die(sprintf("XML error: %s at line %d",
  87                   xml_error_string(xml_get_error_code($xml_parser)),
  88                   xml_get_current_line_number($xml_parser)));
  89           }
  90        }
  91        xml_parser_free($xml_parser);
  92
  93        // return the remaining root element we copied in the beginning
  94        return $ancStack[0];
  95    }
  96
  97 }
  98
  99 $w = new xml2php;
 100 $filename = 'sample.xml';
 101 $result = $w->scanFile( $filename );
 102 $result->myPrint();
 103
 104 return 0;
 105
 106 ?>