includes/ParserXML.php

   1 <?
   2 /**
   3  * This should one day become the XML->(X)HTML parser
   4  * Based on work by Jan Hidders and Magnus Manske
   5  * @package MediaWiki
   6  */
   7
   8 /**
   9  * the base class for an element
  10  * @package MediaWiki
  11  */
  12 class element {
  13   var $name = '';
  14   var $attrs = array();
  15   var $children = array();
  16
  17   function myPrint() {
  18     echo '<UL>';
  19     echo "<LI> <B> Name: </B> $this->name";
  20     // print attributes
  21     echo '<LI> <B> Attributes: </B>';
  22     foreach ($this->attrs as $name => $value) {
  23       echo "$name => $value; " ;
  24     }
  25     // print children
  26     foreach ($this->children as $child) {
  27        if ( is_string($child) ) {
  28          echo '<LI> '.$child;
  29        } else {
  30          $child->myPrint();
  31        }
  32     }
  33     echo '</UL>';
  34   }
  35
  36 }
  37
  38 $ancStack = array();    // the stack with ancestral elements
  39
  40 // Three global functions needed for parsing, sorry guys
  41 function wgXMLstartElement($parser, $name, $attrs) {
  42    global $ancStack, $rootElem;
  43
  44    $newElem = new element;
  45    $newElem->name = $name;
  46    $newElem->attrs = $attrs;
  47    array_push($ancStack, $newElem);
  48    // add to parent if parent exists
  49    $nrAncs = count($ancStack)-1;
  50    if ( $nrAncs > 0 ) {
  51      array_push($ancStack[$nrAncs-1]->children, &$ancStack[$nrAncs]);
  52    } else {
  53      // make extra copy of root element and alias it with the original
  54      array_push($ancStack, &$ancStack[0]);
  55    }
  56 }
  57
  58 function wgXMLendElement($parser, $name) {
  59   global $ancStack, $rootElem;
  60
  61   // pop element of stack
  62   array_pop($ancStack);
  63 }
  64
  65 function wgXMLcharacterData($parser, $data) {
  66   global $ancStack, $rootElem;
  67
  68   $data = trim ( $data ) ; // Don't add blank lines, they're no use...
  69
  70   // add to parent if parent exists
  71   if ( $ancStack && $data != "" ) {
  72     array_push($ancStack[count($ancStack)-1]->children, $data);
  73   }
  74 }
  75
  76
  77 /**
  78  * Here's the class that generates a nice tree
  79  * package parserxml
  80  * @package MediaWiki
  81  */
  82 class xml2php {
  83
  84    function &scanFile( $filename ) {
  85        global $ancStack;
  86        $ancStack = array();
  87
  88        $xml_parser = xml_parser_create();
  89        xml_set_element_handler($xml_parser, 'wgXMLstartElement', 'wgXMLendElement');
  90        xml_set_character_data_handler($xml_parser, 'wgXMLcharacterData');
  91        if (!($fp = fopen($filename, 'r'))) {
  92           die('could not open XML input');
  93        }
  94        while ($data = fread($fp, 4096)) {
  95           if (!xml_parse($xml_parser, $data, feof($fp))) {
  96               die(sprintf("XML error: %s at line %d",
  97                   xml_error_string(xml_get_error_code($xml_parser)),
  98                   xml_get_current_line_number($xml_parser)));
  99           }
 100        }
 101        xml_parser_free($xml_parser);
 102
 103        // return the remaining root element we copied in the beginning
 104        return $ancStack[0];
 105    }
 106
 107 }
 108
 109 $w = new xml2php;
 110 $filename = 'sample.xml';
 111 $result = $w->scanFile( $filename );
 112 $result->myPrint();
 113
 114 return 0;
 115
 116 ?>