xml_domit_lite_parser.php

Joomla!是一套获得过多个奖项的内容管理系统(Content Management System, CMS)。Joomla!采用PHP+MySQL数据库开发

	/**
	* Sets the value of the specified attribute; creates a new attribute if one doesn't exist
	* @param string The attribute name
	* @param string The desired attribute value
	*/
	function setAttribute($name, $value) {
		$this->attributes[$name] = $value;
	} //setAttribute

	/**
	* Removes the specified attribute
	* @param string The name of the attribute to be removed
	*/
	function removeAttribute($name) {
		if ($this->hasAttribute($name)) {
			unset($this->attributes[$name]);
		}
	} //removeAttribute

	/**
	* Determines whether an attribute with the specified name exists
	* @param string The name of the attribute
	* @return boolean True if the attribute exists
	*/
	function hasAttribute($name) {
		return isset($this->attributes[$name]);
	} //hasAttribute

	/**
	* Collapses adjacent text nodes in entire element subtree
	*/
	function normalize() {
		if ($this->hasChildNodes()) {
			$currNode =& $this->childNodes[0];

			while ($currNode->nextSibling != null) {
				$nextNode =& $currNode->nextSibling;

				if (($currNode->nodeType == DOMIT_TEXT_NODE) &&
					($nextNode->nodeType == DOMIT_TEXT_NODE)) {
					$currNode->nodeValue .= $nextNode->nodeValue;
					$this->removeChild($nextNode);
				}
				else {
					$currNode->normalize();
				}

				if ($currNode->nextSibling != null) {
					$currNode =& $currNode->nextSibling;
				}
			}
		}
	} //normalize

	/**
	* Generates an array representation of the node and its children
	* @return Array A representation of the node and its children
	*/
	function toArray() {
		$arReturn = array($this->nodeName => array("attributes" => $this->attributes));
		$total = $this->childCount;

		for ($i = 0; $i < $total; $i++) {
			$arReturn[$this->nodeName][$i] = $this->childNodes[$i]->toArray();
		}

		return $arReturn;
	} //toArray

	/**
	* Copies a node and/or its children
	* @param boolean True if all child nodes are also to be cloned
	* @return Object A copy of the node and/or its children
	*/
	function &cloneNode($deep = false) {
		$className = get_class($this);
		$clone = new $className($this->nodeName);

		$clone->attributes = $this->attributes;

		if ($deep) {
			$total = $this->childCount;

			for ($i = 0; $i < $total; $i++) {
				$currentChild =& $this->childNodes[$i];
				$clone->appendChild($currentChild->cloneNode($deep));
			}
		}

		return $clone;
	} //cloneNode

	/**
	* Generates a string representation of the node and its children
	* @param boolean True if HTML readable output is desired
	* @param boolean True if illegal xml characters in text nodes and attributes should be converted to entities
	* @return string The string representation
	*/
	function toString($htmlSafe = false, $subEntities = false) {
		require_once(DOMIT_INCLUDE_PATH . 'xml_domit_utilities.php');
		global $DOMIT_defined_entities_flip;

		$result = '<' . $this->nodeName;

		//get attributes
		foreach ($this->attributes as $key => $value) {
			$result .= ' ' . $key . '="';
			$result .= $subEntities ? DOMIT_Utilities::convertEntities($value,
							$DOMIT_defined_entities_flip) : $value;
			$result .= '"';
		}

		//get children
		$myNodes =& $this->childNodes;
		$total = count($myNodes);

		if ($total != 0) {
			$result .= '>';

			for ($i = 0; $i < $total; $i++) {
				$child =& $myNodes[$i];
				$result .= $child->toString(false, $subEntities);
			}

			$result .= '</' . $this->nodeName . '>';
		}
		else {
			if ($this->ownerDocument->doExpandEmptyElementTags) {
				if (in_array($this->nodeName, $this->ownerDocument->expandEmptyElementExceptions)) {
					$result .= ' />';
				}
				else {
                	$result .= '></' . $this->nodeName . '>';
				}
		    }
		    else {
				if (in_array($this->nodeName, $this->ownerDocument->expandEmptyElementExceptions)) {
					$result .= '></' . $this->nodeName . '>';
				}
				else {
					$result .= ' />';
				}
		    }
		}

		if ($htmlSafe) $result = $this->forHTML($result);

		return $result;
	} //toString
} //DOMIT_Lite_Element

/**
* A class representing the DOM Text Node
*
* @package domit-xmlparser
* @subpackage domit-xmlparser-lite
* @author John Heinstein <[e-mail]>
*/
class DOMIT_Lite_TextNode extends DOMIT_Lite_Node {
	/**
	* DOM Text Node constructor
	* @param string The text of the node
	*/
	function DOMIT_Lite_TextNode($data) {
		$this->_constructor();
		$this->nodeType = DOMIT_TEXT_NODE;
		$this->nodeName = '#text';
		$this->setText($data);
	} //DOMIT_Lite_TextNode

	/**
	* Returns the text contained in the current node
	* @return string The text of the current node
	*/
	function getText() {
		return $this->nodeValue;
	} //getText

 	/**
	* Sets the text contained in the current node to $data.
	* @param string The text data of the node
	*/
	function setText($data) {
        $this->nodeValue = $data;
	} //setText

	/**
	* Generates an array representation of the node and its children
	* @return Array A representation of the node and its children
	*/
	function toArray() {
		return $this->toString();
	} //toArray

	/**
	* Copies a node and/or its children
	* @param boolean True if all child nodes are also to be cloned
	* @return Object A copy of the node and/or its children
	*/
	function &cloneNode($deep = false) {
		$className = get_class($this);
		$clone = new $className($this->nodeValue);

		return $clone;
	} //cloneNode

	/**
	* Generates a string representation of the node and its children
	* @param boolean True if HTML readable output is desired
	* @param boolean True if illegal xml characters should be converted to entities
	* @return string The string representation
	*/
	function toString($htmlSafe = false, $subEntities = false) {
		require_once(DOMIT_INCLUDE_PATH . 'xml_domit_utilities.php');
		global $DOMIT_defined_entities_flip;

		$result = $subEntities ? DOMIT_Utilities::convertEntities($this->nodeValue,
						$DOMIT_defined_entities_flip) : $this->nodeValue;

		if ($htmlSafe) $result = $this->forHTML($result);

		return $result;
	} //toString
} //DOMIT_Lite_TextNode

/**
* A class representing the DOM CDATA Section
*
* @package domit-xmlparser
* @subpackage domit-xmlparser-lite
* @author John Heinstein <[e-mail]>
*/
class DOMIT_Lite_CDATASection extends DOMIT_Lite_TextNode {
	/**
	* DOM CDATA Section node constructor
	* @param string The text of the node
	*/
	function DOMIT_Lite_CDATASection($data) {
		$this->_constructor();
		$this->nodeType = DOMIT_CDATA_SECTION_NODE;
		$this->nodeName = '#cdata-section';
		$this->setText($data);
	} //DOMIT_Lite_CDATASection

	/**
	* Generates a string representation of the node and its children
	* @param boolean True if HTML readable output is desired
	* @param boolean True if illegal xml characters should be converted to entities
	* @return string The string representation
	*/
	function toString($htmlSafe = false, $subEntities = false) {
		$result = '<![CDATA[';
		$result .= $subEntities ? str_replace("]]>", "]]&gt;", $this->nodeValue) :
								 $this->nodeValue;
		$result .= ']]>';

		if ($htmlSafe) $result = $this->forHTML($result);

		return $result;
	} //toString
} //DOMIT_Lite_CDATASection

/**
* Manages the generation of a DOMIT! document from SAX events
*
* @package domit-xmlparser
* @subpackage domit-xmlparser-lite
* @author John Heinstein <[e-mail]>
*/
class DOMIT_Lite_Parser {
	/** @var Object A reference to the resulting xmldoc */
	var $xmlDoc = null;
	/** @var Object A reference to the current node in the parsing process */
	var $currentNode = null;
	/** @var Object A reference to the last child in the parsing process */
	var $lastChild = null;
	/** @var boolean True if currently parsing a CDATA Section */
	var $inCDATASection = false; //flag for Expat
	/** @var boolean True if currently parsing a Text node */
	var $inTextNode = false;
	/** @var boolean True is CDATA Section nodes are not to be converted into Text nodes */
	var $preserveCDATA;
	/** @var string A container for holding the currently parsed text data */
	var $parseContainer = '';
	/** @var string The current docutype text */
	var $parseItem = '';

	/**
	* Parses xml text using Expat
	* @param Object A reference to the DOM document that the xml is to be parsed into
	* @param string The text to be parsed
	* @param boolean True if CDATA Section nodes are not to be converted into Text nodes
	* @return boolean True if the parsing is successful
	*/
	function parse (&$myXMLDoc, $xmlText, $preserveCDATA = true) {
		$this->xmlDoc =& $myXMLDoc;
		$this->lastChild =& $this->xmlDoc;

		$this->preserveCDATA = $preserveCDATA;

		//create instance of expat parser (should be included in php distro)
		if (version_compare(phpversion(), '5.0', '<=')) {
			$parser = xml_parser_create('');
		}
		else {
			$parser = xml_parser_create();
		}

		//set handlers for SAX events
		xml_set_object($parser, $this);
		xml_set_element_handler($parser, 'startElement', 'endElement');
		xml_set_character_data_handler($parser, 'dataElement');
		xml_set_default_handler($parser, 'defaultDataElement');
		xml_parser_set_option($parser, XML_OPTION_CASE_FOLDING, 0);

		if (!$this->xmlDoc->preserveWhitespace) {
			xml_parser_set_option($parser, XML_OPTION_SKIP_WHITE, 1);
		}
		else {
			xml_parser_set_option($parser, XML_OPTION_SKIP_WHITE, 0);
		}

		//parse out whitespace -  (XML_OPTION_SKIP_WHITE = 1 does not
		//seem to work consistently across versions of PHP and Expat
		if (!$this->xmlDoc->preserveWhitespace) {
			$xmlText = eregi_replace('>' . "[[:space:]]+" . '<' , '><', $xmlText);
		}

		$success = xml_parse($parser, $xmlText);

		$this->xmlDoc->errorCode = xml_get_error_code($parser);
		$this->xmlDoc->errorString = xml_error_string($this->xmlDoc->errorCode);

		xml_parser_free($parser);

		return $success;
	} //parse

	/**
	* Parses xml text using SAXY
	* @param Object A reference to the DOM document that the xml is to be parsed into
	* @param string The text to be parsed
	* @param boolean True if CDATA Section nodes are not to be converted into Text nodes
	* @return boolean True if the parsing is successful
	*/
	function parseSAXY(&$myXMLDoc, $xmlText, $preserveCDATA, $definedEntities) {
		require_once(DOMIT_INCLUDE_PATH . 'xml_saxy_lite_parser.php');

		$this->xmlDoc =& $myXMLDoc;
		$this->lastChild =& $this->xmlDoc;

		//create instance of SAXY parser
		$parser = new SAXY_Lite_Parser();
		$parser->appendEntityTranslationTable($definedEntities);

		//not yet implemented in SAXY!!!
		$parser->preserveWhitespace = $this->xmlDoc->preserveWhitespace;

		$parser->xml_set_element_handler(array(&$this, 'startElement'), array(&$this, 'endElement'));
		$parser->xml_set_character_data_handler(array(&$this, 'dataElement'));

		if ($preserveCDATA) {
			$parser->xml_set_cdata_section_handler(array(&$this, 'cdataElement'));
		}

		$success = $parser->parse($xmlText);

		$this->xmlDoc->errorCode = $parser->xml_get_error_code();
		$this->xmlDoc->errorString = $parser->xml_error_string($this->xmlDoc->errorCode);

		return $success;
	} //parseSAXY

	/**
	* Generates and appends a new text node from the parseContainer text
	*/
	function dumpTextNode() {
		//traps for mixed content
	    $currentNode =& $this->xmlDoc->createTextNode($this->parseContainer);
		$this->lastChild->appendChild($currentNode);
		$this->inTextNode = false;
		$this->parseContainer = '';
	} //dumpTextNode

	/**
	* Catches a start element event and processes the data
	* @param Object A reference to the current SAX parser
	* @param string The tag name of the current element
	* @param Array An array of the element attributes
	*/
	function startElement(&$parser, $name, $attrs) {
		if ($this->inTextNode) {
			$this->dumpTextNode();
		}

		$currentNode =& $this->xmlDoc->createElement($name);
		$currentNode->attributes = $attrs;
		$this->lastChild->appendChild($currentNode);
		$this->lastChild =& $currentNode;
	} //startElement

	/**
	* Catches an end element event and processes the data
	* @param Object A reference to the current SAX parser
	* @param string The tag name of the current element
	*/
	function endElement(&$parser, $name) {
		if ($this->inTextNode) {
			$this->dumpTextNode();
		}

		$this->lastChild =& $this->lastChild->parentNode;
	} //endElement

	/**
	* Catches a data event and processes the text
	* @param Object A reference to the current SAX parser
	* @param string The current text data
	*/
	function dataElement(&$parser, $data) {
		if (!$this->inCDATASection) $this->inTextNode = true;

		$this->parseContainer .= $data;
	} //dataElement

	/**
	* Catches a CDATA Section event and processes the text
	* @param Object A reference to the current SAX parser
	* @param string The current text data
	*/
	function cdataElement(&$parser, $data) {
		$currentNode =& $this->xmlDoc->createCDATASection($data);

		$this->lastChild->appendChild($currentNode);
	} //cdataElement

	/**
	* Catches a default data event and processes the data
	* @param Object A reference to the current SAX parser
	* @param string The current data
	*/
	function defaultDataElement(&$parser, $data) {
		if (strlen($data) > 2){
			$pre = strtoupper(substr($data, 0, 3));

			switch ($pre) {
				case '<![': //cdata section coming
					if ($this->preserveCDATA) {
						$this->inCDATASection = true;
					}
					break;
				case ']]>': //cdata remnant - ignore
					if ($this->preserveCDATA) {
						$currentNode =& $this->xmlDoc->createCDATASection($this->parseContainer);
						$this->lastChild->appendChild($currentNode);
						$this->inCDATASection = false;
						$this->parseContainer = '';
					}
					else {
						$this->dumpTextNode();
					}

					break;
			}
		}
	} //defaultDataElement
} //DOMIT_Lite_Parser

?>