xml_saxy_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

<?php
/**
* SAXY is a non-validating, but lightweight and fast SAX parser for PHP, modelled on the Expat parser
* @package saxy-xmlparser
* @subpackage saxy-xmlparser-main
* @version 1.0
* @copyright (C) 2004 John Heinstein. All rights reserved
* @license http://www.gnu.org/copyleft/lesser.html LGPL License
* @author John Heinstein <[e-mail]>
* @link http://www.engageinteractive.com/saxy/ SAXY Home Page
* SAXY is Free Software
**/

if (!defined('SAXY_INCLUDE_PATH')) {
	define('SAXY_INCLUDE_PATH', (dirname(__FILE__) . "/"));
}

/** current version of SAXY */
define ('SAXY_VERSION', '1.0');

/** default XML namespace */
define ('SAXY_XML_NAMESPACE', 'http://www.w3.org/xml/1998/namespace');

/** saxy parse state, before prolog is encountered */
define('SAXY_STATE_PROLOG_NONE', 0);
/** saxy parse state, in processing instruction */
define('SAXY_STATE_PROLOG_PROCESSINGINSTRUCTION', 1);
/** saxy parse state, an exclamation mark has been encountered */
define('SAXY_STATE_PROLOG_EXCLAMATION', 2);
/** saxy parse state, in DTD */
define('SAXY_STATE_PROLOG_DTD', 3);
/** saxy parse state, an inline DTD */
define('SAXY_STATE_PROLOG_INLINEDTD', 4);
/** saxy parse state, a comment */
define('SAXY_STATE_PROLOG_COMMENT', 5);
/** saxy parse state, processing main document */
define('SAXY_STATE_PARSING', 6);
/** saxy parse state, processing comment in main document */
define('SAXY_STATE_PARSING_COMMENT', 7);

//SAXY error codes; same as EXPAT error codes
/** no error */
define('SAXY_XML_ERROR_NONE', 0);
/** out of memory error */
define('SAXY_XML_ERROR_NO_MEMORY', 1);
/** syntax error */
define('SAXY_XML_ERROR_SYNTAX', 2);
/** no elements in document */
define('SAXY_XML_ERROR_NO_ELEMENTS', 3);
/** invalid token encountered error */
define('SAXY_XML_ERROR_INVALID_TOKEN', 4);
/** unclosed token error */
define('SAXY_XML_ERROR_UNCLOSED_TOKEN', 5);
/** partial character error */
define('SAXY_XML_ERROR_PARTIAL_CHAR', 6);
/** mismatched tag error */
define('SAXY_XML_ERROR_TAG_MISMATCH', 7);
/** duplicate attribute error */
define('SAXY_XML_ERROR_DUPLICATE_ATTRIBUTE', 8);
/** junk after document element error */
define('SAXY_XML_ERROR_JUNK_AFTER_DOC_ELEMENT', 9);
/** parameter enitity reference error */
define('SAXY_XML_ERROR_PARAM_ENTITY_REF', 10);
/** undefined entity error */
define('SAXY_XML_ERROR_UNDEFINED_ENTITY', 11);
/** recursive entity error */
define('SAXY_XML_ERROR_RECURSIVE_ENTITY_REF', 12);
/** asynchronous entity error */
define('SAXY_XML_ERROR_ASYNC_ENTITY', 13);
/** bad character reference error */
define('SAXY_XML_ERROR_BAD_CHAR_REF', 14);
/** binary entity reference error */
define('SAXY_XML_ERROR_BINARY_ENTITY_REF', 15);
/** attribute external entity error */
define('SAXY_XML_ERROR_ATTRIBUTE_EXTERNAL_ENTITY_REF', 16);
/** misplaced processing instruction error */
define('SAXY_XML_ERROR_MISPLACED_XML_PI', 17);
/** unknown encoding error */
define('SAXY_XML_ERROR_UNKNOWN_ENCODING', 18);
/** incorrect encoding error */
define('SAXY_XML_ERROR_INCORRECT_ENCODING', 19);
/** unclosed CDATA Section error */
define('SAXY_XML_ERROR_UNCLOSED_CDATA_SECTION', 20);
/** external entity handling error */
define('SAXY_XML_ERROR_EXTERNAL_ENTITY_HANDLING', 21);

require_once(SAXY_INCLUDE_PATH . 'xml_saxy_shared.php');

/**
* The SAX Parser class
*
* @package saxy-xmlparser
* @subpackage saxy-xmlparser-main
* @author John Heinstein <[e-mail]>
*/
class SAXY_Parser extends SAXY_Parser_Base {
    /** @var int The current error number */
	var $errorCode = SAXY_XML_ERROR_NONE;
	/** @var Object A reference to the DocType event handler */
	var $DTDHandler = null;
	/** @var Object A reference to the Comment event handler */
	var $commentHandler = null;
	/** @var Object A reference to the Processing Instruction event handler */
	var $processingInstructionHandler = null;
	/** @var Object A reference to the Start Namespace Declaration event handler */
	var $startNamespaceDeclarationHandler = null;
	/** @var Object A reference to the End Namespace Declaration event handler */
	var $endNamespaceDeclarationHandler = null;
	/** @var boolean True if SAXY takes namespaces into consideration when parsing element tags */
	var $isNamespaceAware = false;
	/** @var array An indexed array containing associative arrays of namespace prefixes mapped to their namespace URIs */
	var $namespaceMap = array();
	/** @var array A stack used to determine when an end namespace event should be fired */
	var $namespaceStack = array();
	/** @var array A track used to track the uri of the current default namespace */
	var $defaultNamespaceStack = array();
	/** @var array A stack containing tag names of unclosed elements */
	var $elementNameStack = array();

	/**
	* Constructor for SAX parser
	*/
	function SAXY_Parser() {
		$this->SAXY_Parser_Base();
		$this->state = SAXY_STATE_PROLOG_NONE;
	} //SAXY_Parser

	/**
	* Sets a reference to the handler for the DocType event
	* @param mixed A reference to the DocType handler
	*/
	function xml_set_doctype_handler($handler) {
		$this->DTDHandler =& $handler;
	} //xml_set_doctype_handler

	/**
	* Sets a reference to the handler for the Comment event
	* @param mixed A reference to the Comment handler
	*/
	function xml_set_comment_handler($handler) {
		$this->commentHandler =& $handler;
	} //xml_set_comment_handler

	/**
	* Sets a reference to the handler for the Processing Instruction event
	* @param mixed A reference to the Processing Instruction handler
	*/
	function xml_set_processing_instruction_handler($handler) {
		$this->processingInstructionHandler =& $handler;
	} //xml_set_processing_instruction_handler

	/**
	* Sets a reference to the handler for the Start Namespace Declaration event
	* @param mixed A reference to the Start Namespace Declaration handler
	*/
	function xml_set_start_namespace_decl_handler($handler) {
		$this->startNamespaceDeclarationHandler =& $handler;
	} //xml_set_start_namespace_decl_handler

	/**
	* Sets a reference to the handler for the End Namespace Declaration event
	* @param mixed A reference to the Start Namespace Declaration handler
	*/
	function xml_set_end_namespace_decl_handler($handler) {
		$this->endNamespaceDeclarationHandler =& $handler;
	} //xml_set_end_namespace_decl_handler

	/**
	* Specifies whether SAXY is namespace sensitive
	* @param boolean True if SAXY is namespace aware
	*/
	function setNamespaceAwareness($isNamespaceAware) {
		$this->isNamespaceAware =& $isNamespaceAware;
	} //setNamespaceAwareness

	/**
	* Returns the current version of SAXY
	* @return Object The current version of SAXY
	*/
	function getVersion() {
		return SAXY_VERSION;
	} //getVersion

	/**
	* Processes the xml prolog, doctype, and any other nodes that exist outside of the main xml document
	* @param string The xml text to be processed
	* @return string The preprocessed xml text
	*/
	function preprocessXML($xmlText) {
		//strip prolog
		$xmlText = trim($xmlText);
		$startChar = -1;
		$total = strlen($xmlText);

		for ($i = 0; $i < $total; $i++) {
//			$currentChar = $xmlText{$i};
			$currentChar = substr($xmlText, $i, 1);

			switch ($this->state) {
				case SAXY_STATE_PROLOG_NONE:
					if ($currentChar == '<') {
						$nextChar = $xmlText{($i + 1)};

						if ($nextChar == '?')  {
							$this->state = SAXY_STATE_PROLOG_PROCESSINGINSTRUCTION;
							$this->charContainer = '';
						}
						else if ($nextChar == '!') {
							$this->state = SAXY_STATE_PROLOG_EXCLAMATION;
							$this->charContainer .= $currentChar;
							break;
						}
						else {
							$this->charContainer = '';
							$startChar  = $i;
							$this->state = SAXY_STATE_PARSING;
							return (substr($xmlText, $startChar));
						}
					}

					break;

				case SAXY_STATE_PROLOG_EXCLAMATION:
					if ($currentChar == 'D') {
						$this->state = SAXY_STATE_PROLOG_DTD;
						$this->charContainer .= $currentChar;
					}
					else if ($currentChar == '-') {
						$this->state = SAXY_STATE_PROLOG_COMMENT;
						$this->charContainer = '';
					}
					else {
						//will trap ! and add it
						$this->charContainer .= $currentChar;
					}

					break;

				case SAXY_STATE_PROLOG_PROCESSINGINSTRUCTION:
					if ($currentChar == '>') {
						$this->state = SAXY_STATE_PROLOG_NONE;
						$this->parseProcessingInstruction($this->charContainer);
						$this->charContainer = '';
					}
					else {
						$this->charContainer .= $currentChar;
					}

					break;

				case SAXY_STATE_PROLOG_COMMENT:
					if ($currentChar == '>') {
						$this->state = SAXY_STATE_PROLOG_NONE;
						$this->parseComment($this->charContainer);
						$this->charContainer = '';
					}
					else if ($currentChar == '-') {
						if ((($xmlText{($i + 1)} == '-')  && ($xmlText{($i + 2)} == '>')) ||
							($xmlText{($i + 1)} == '>') ||
							(($xmlText{($i - 1)} == '-')  && ($xmlText{($i - 2)}== '!')) ){
							//do nothing
						}
						else {
							$this->charContainer .= $currentChar;
						}
					}
					else {
						$this->charContainer .= $currentChar;
					}

					break;

				case SAXY_STATE_PROLOG_DTD:
					if ($currentChar == '[') {
						$this->charContainer .= $currentChar;
						$this->state = SAXY_STATE_PROLOG_INLINEDTD;
					}
					else if ($currentChar == '>') {
						$this->state = SAXY_STATE_PROLOG_NONE;

						if ($this->DTDHandler != null) {
							$this->fireDTDEvent($this->charContainer . $currentChar);
						}

						$this->charContainer = '';
					}
					else {
						$this->charContainer .= $currentChar;
					}

					break;

				case SAXY_STATE_PROLOG_INLINEDTD:
					$previousChar = $xmlText{($i - 1)};

					if (($currentChar == '>') && ($previousChar == ']')){
						$this->state = SAXY_STATE_PROLOG_NONE;

						if ($this->DTDHandler != null) {
							$this->fireDTDEvent($this->charContainer . $currentChar);
						}

						$this->charContainer = '';
					}
					else {
						$this->charContainer .= $currentChar;
					}

					break;

			}
		}
	} //preprocessXML

	/**
	* The controlling method for the parsing process
	* @param string The xml text to be processed
	* @return boolean True if parsing is successful
	*/
	function parse ($xmlText) {
		$xmlText = $this->preprocessXML($xmlText);
		$total = strlen($xmlText);

		for ($i = 0; $i < $total; $i++) {
//			$currentChar = $xmlText{$i};
			$currentChar = substr($xmlText, $i, 1);

			switch ($this->state) {
				case SAXY_STATE_PARSING:
					switch ($currentChar) {
						case '<':
							if (substr($this->charContainer, 0, SAXY_CDATA_LEN) == SAXY_SEARCH_CDATA) {
								$this->charContainer .= $currentChar;
							}
							else {
								$this->parseBetweenTags($this->charContainer);
								$this->charContainer = '';
							}
							break;

						case '-':
							if (($xmlText{($i - 1)} == '-') && ($xmlText{($i - 2)} == '!')
								&& ($xmlText{($i - 3)} == '<')) {
								$this->state = SAXY_STATE_PARSING_COMMENT;
								$this->charContainer = '';
							}
							else {
								$this->charContainer .= $currentChar;
							}
							break;

						case '>':
							if ((substr($this->charContainer, 0, SAXY_CDATA_LEN) == SAXY_SEARCH_CDATA) &&
								!(($this->getCharFromEnd($this->charContainer, 0) == ']') &&
								($this->getCharFromEnd($this->charContainer, 1) == ']'))) {
								$this->charContainer .= $currentChar;
							}
							else {
								$this->parseTag($this->charContainer);
								$this->charContainer = '';
							}
							break;

						default:
							$this->charContainer .= $currentChar;
					}

					break;

				case SAXY_STATE_PARSING_COMMENT:
					switch ($currentChar) {
						case '>':
							if (($xmlText{($i - 1)} == '-') && ($xmlText{($i - 2)} == '-')) {
								$this->fireCommentEvent(substr($this->charContainer, 0,
													(strlen($this->charContainer) - 2)));
								$this->charContainer = '';
								$this->state = SAXY_STATE_PARSING;
							}
							else {
								$this->charContainer .= $currentChar;
							}
							break;

						default:
							$this->charContainer .= $currentChar;
					}

					break;
			}
		}

		return ($this->errorCode == 0);
	} //parse

	/**
	* Parses an element tag
	* @param string The interior text of the element tag
	*/
	function parseTag($tagText) {
		$tagText = trim($tagText);
		$firstChar = $tagText{0};
		$myAttributes = array();

		switch ($firstChar) {
			case '/':
				$tagName = substr($tagText, 1);
				$this->_fireEndElementEvent($tagName);
				break;

			case '!':
				$upperCaseTagText = strtoupper($tagText);

				if (strpos($upperCaseTagText, SAXY_SEARCH_CDATA) !== false) { //CDATA Section
					$total = strlen($tagText);
					$openBraceCount = 0;
					$textNodeText = '';

					for ($i = 0; $i < $total; $i++) {
//						$currentChar = $tagText{$i};
						$currentChar = substr($tagText, $i, 1);

						if (($currentChar == ']') && ($tagText{($i + 1)} == ']')) {
							break;
						}