<?php


# This script formats a xml file according to some rules. These rules 
# are processed in the function that xml_set_character_data_handler 
# defines. We have to remember, in which element we are, and furthermore, 
# which ancestor elements the current element has. 

/** gm 04/10/03 - Made changes:
	- renamed to ConfigParse
	- added ConfigParse::$error, and changed ConfigParse::ConfigParse() to use that instead of die()
	- No longer addeds node for ending element  ( </element> )
	- removed fancy printing crap
	- added method ConfigParse::getElement() and ConfigParse::getElements()
	- bugfix- added ConfigParse::$_lastEnd, to not add cdata right after a </element>
	  this means that in the case of: ... </element1> CDATA </element2> the CDATA will
	  be ignored.  Note: this could also be changed to add the CDATA to <element2>'s node
	- changed Node::add_data so it doesn't always add a space at the start
	- removed Node::$type variable
	- added ConfigParse::getElementData and ConfigParse:getElementAttrib
	
	- renamed back to XML()
	- pass xml data instead of filename
*/


class Node { 
    var $name; 
    var $attributes; 
    var $ancestors = "/";
    var $data; 

    function Node($tree) { 
        $this->name = array_pop($tree); 
        $this->ancestors .= implode("/", $tree); 
    } 

    function add_data($value) { 
        $this->data = (strlen($this->data) > 0 ? $this->data.' '.$value : $value); 
    } 

    function get_type() { 
        if (strlen($this->data) > 0) { 
            return "with CDATA";
        } else { 
            return "without CDATA";
        } 
    } 

    function level() { 
        if ($this->ancestors == "/") return 0; 
      if (preg_match_all("/(\/{1})/", $this->ancestors, $result,PREG_PATTERN_ORDER)) { 
        return (count($result[0])); 
      } else { 
        return 0;
        }
    }

    function has_attributes() {
        return (is_array($this->attributes));
    }

    function print_name() {
        return "$this->name";
    }

    function is_child($node) {
        $result = preg_match("/^$ancestors/", $node->ancestors, $match);
        if ($node->ancestors == $this->ancestors) $result = false;
        return $result;
    }
}

class XML {
    var $tree = array();
    var $nodes = array();
    var $PIs;
    var $error;
    var $_lastEnd;

    function XML($data) {
	$this->_error = false;
        $xml_parser = xml_parser_create();
        xml_set_object($xml_parser,$this);
        xml_set_element_handler($xml_parser, "startElement", "endElement");
        xml_set_character_data_handler($xml_parser, "characterData");
        xml_set_processing_instruction_handler ($xml_parser, "process_instruction");
                # Why should one want to use case-folding with XML? XML is case-sensitiv, I think this is nonsense
        xml_parser_set_option($xml_parser, XML_OPTION_CASE_FOLDING, false);

        if (!xml_parse($xml_parser, $data, true)) {
             $this->error = sprintf("XML error: %s at line %d",
                                    xml_error_string(xml_get_error_code($xml_parser)),
                                    xml_get_current_line_number($xml_parser));
	}
    }

    function startElement($parser, $name, $attribs) { 
        # Adding the additional element to the tree, including attributes 
        $this->tree[] = $name; 

        $node = new Node($this->tree); 
        while (list($k, $v) = each($attribs)) { 
            $node->attributes[$k] = $v; 
        } 
        $this->nodes[] = $node;
	$this->_lastEnd = false; 
    } 

    function endElement($parser, $name) { 
        # Adding a new element, describing the end of the tag 
        # But only, if the Tag has CDATA in it! 
         
        # Check 
        if (count($this->nodes) >= 1) { 
            $prev_node = $this->nodes[count($this->nodes)-1]; 
            if (strlen($prev_node->data) > 0 || $prev_node->name != $name) {
                $this->tree[count($this->tree)-1] = "/".$this->tree[count($this->tree)-1];
		 
		// gm 04/10/03 - dont add element for end node
                //$this->nodes[] = new Node($this->tree, NULL); 
            } else { 
                # Adding a slash to the end of the prev_node 
                $prev_node->name = $prev_node->name."/";
                $this->nodes[count($this->nodes)-1]->name = $this->nodes[count($this->nodes)-1]->name."/" ;
            } 
        } 

        # Removing the element from the tree 
        array_pop($this->tree); 
	$this->_lastEnd = true; 
    } 

    function characterData($parser, $data) { 
        $data = ltrim($data); 
        if (($data != "") && (!$this->_lastEnd)) {
	    $this->nodes[count($this->nodes)-1]->add_data($data);
	} 
    } 
     
    function process_instruction($parser, $target, $data) { 
        if (preg_match("/xml:stylesheet/", $target, $match) && preg_match("/type=\"text\/xsl\"/", $data, $match)) { 
            preg_match("/href=\"(.+)\"/i", $data, $this->PIs); 
#            print "<b>found xls pi: $PIs[1]</b><br>\n" 
        } 
    } 
    
    // gets the element specified by path (first match)
    // path is in the format "/Container/Of/Element"
    // if the element is single (ie, "<something />") then include a trailing / ("/Container/Of/Element/")
    function &getElement($path) {
    	// see if this is a single element (ie, <something /> ), if so, remove
	// trailing / so as not to mess up explode() etc
    	$slash = false;
    	if ($path[strlen($path)-1] == "/") {
		$path = substr($path, 0, -1);
		$slash = true;
	}
	
    	// change into "/Container/Of" and "Element"
    	$temp = explode("/", $path);
	$name = array_pop($temp).($slash ? "/" : ""); // add slash back, if it needs it
	$path = implode("/", $temp);
    	
        foreach ($this->nodes as $node) {
            if ($node->ancestors == $path) {
		if ($node->name == $name) {
		        return $node;
		}
	    }
	}
	return false;
    }
    
    // get the CDATA for an element
    function getElementData($path) {
	$node = $this->getElement($path);
	if ($node) {
		return $node->data;
	}
    }
    
    // get a specific attribute for an element
    // NOTE: if calling this multiple times for the same element, its better to use getElement()
    // and access the attributes directly (otherwise you're seaching through the nodes many times)
    function getElementAttrib($path, $attrib) {
	$node = $this->getElement($path);
	if (isset($node->attributes[$attrib])) {
		return $node->attributes[$attrib];
	}
	return false;
    }

    // gets the elements specified by path (into an array)
    // path is in the format "/Container/Of/Element"
    function getElements($path) {
    	$return = array();
	
        foreach ($this->nodes as $node) {
            if ($node->ancestors == $path) {
	        $return[] = $node;
	    }
	}
	
	return $return;
    }

} 
?>