Original port from Python | // | Authors: Harry Fuecks Port to PEAR + more | // | Authors: Many @ Sitepointforums Advanced PHP Forums | // +----------------------------------------------------------------------+ // // $Id: HTMLSax3.php 3188 2012-07-12 12:13:23Z ctrlaltca $ // /** * Main parser components * @package System.Security.SafeHtml * @version $Id: HTMLSax3.php 3188 2012-07-12 12:13:23Z ctrlaltca $ */ /** * Required classes */ require_once(dirname(__FILE__).'/HTMLSax3/States.php'); require_once(dirname(__FILE__).'/HTMLSax3/Decorators.php'); /** * Base State Parser * @package System.Security.SafeHtml * @access protected * @abstract */ class TSax3_StateParser { /** * Instance of user front end class to be passed to callbacks * @var TSax3 * @access private */ public $htmlsax; /** * User defined object for handling elements * @var object * @access private */ public $handler_object_element; /** * User defined open tag handler method * @var string * @access private */ public $handler_method_opening; /** * User defined close tag handler method * @var string * @access private */ public $handler_method_closing; /** * User defined object for handling data in elements * @var object * @access private */ public $handler_object_data; /** * User defined data handler method * @var string * @access private */ public $handler_method_data; /** * User defined object for handling processing instructions * @var object * @access private */ public $handler_object_pi; /** * User defined processing instruction handler method * @var string * @access private */ public $handler_method_pi; /** * User defined object for handling JSP/ASP tags * @var object * @access private */ public $handler_object_jasp; /** * User defined JSP/ASP handler method * @var string * @access private */ public $handler_method_jasp; /** * User defined object for handling XML escapes * @var object * @access private */ public $handler_object_escape; /** * User defined XML escape handler method * @var string * @access private */ public $handler_method_escape; /** * User defined handler object or NullHandler * @var object * @access private */ public $handler_default; /** * Parser options determining parsing behavior * @var array * @access private */ protected $parser_options = array(); /** * XML document being parsed * @var string * @access private */ protected $rawtext; /** * Position in XML document relative to start (0) * @var int * @access private */ protected $position; /** * Length of the XML document in characters * @var int * @access private */ protected $length; /** * Array of state objects * @var array * @access private */ protected $State = array(); const TSAX3_STATE_STOP = 0; const TSAX3_STATE_START = 1; const TSAX3_STATE_TAG = 2; const TSAX3_STATE_OPENING_TAG = 3; const TSAX3_STATE_CLOSING_TAG = 4; const TSAX3_STATE_ESCAPE = 6; const TSAX3_STATE_JASP = 7; const TSAX3_STATE_PI = 8; /** * Constructs TSax3_StateParser setting up states * @var TSax3 instance of user front end class * @access protected */ protected function __construct($htmlsax) { $this->htmlsax = $htmlsax; $this->State[self::TSAX3_STATE_START] = new TSax3_StartingState(); $this->State[self::TSAX3_STATE_CLOSING_TAG] = new TSax3_ClosingTagState(); $this->State[self::TSAX3_STATE_TAG] = new TSax3_TagState(); $this->State[self::TSAX3_STATE_OPENING_TAG] = new TSax3_OpeningTagState(); $this->State[self::TSAX3_STATE_PI] = new TSax3_PiState(); $this->State[self::TSAX3_STATE_JASP] = new TSax3_JaspState(); $this->State[self::TSAX3_STATE_ESCAPE] = new TSax3_EscapeState(); } /** * Moves the position back one character * @access protected * @return void */ function unscanCharacter() { $this->position -= 1; } /** * Moves the position forward one character * @access protected * @return void */ function ignoreCharacter() { $this->position += 1; } /** * Returns the next character from the XML document or void if at end * @access protected * @return mixed */ function scanCharacter() { if ($this->position < $this->length) { return $this->rawtext{$this->position++}; } } /** * Returns a string from the current position to the next occurance * of the supplied string * @param string string to search until * @access protected * @return string */ function scanUntilString($string) { $start = $this->position; $this->position = strpos($this->rawtext, $string, $start); if ($this->position === FALSE) { $this->position = $this->length; } return substr($this->rawtext, $start, $this->position - $start); } /** * Returns a string from the current position until the first instance of * one of the characters in the supplied string argument * @param string string to search until * @access protected * @return string * @abstract */ function scanUntilCharacters($string) {} /** * Moves the position forward past any whitespace characters * @access protected * @return void * @abstract */ function ignoreWhitespace() {} /** * Begins the parsing operation, setting up any decorators, depending on * parse options invoking _parse() to execute parsing * @param string XML document to parse * @access protected * @return void */ function parse($data) { if ($this->parser_options['XML_OPTION_TRIM_DATA_NODES']==1) { $decorator = new TSax3_Trim( $this->handler_object_data, $this->handler_method_data); $this->handler_object_data =& $decorator; $this->handler_method_data = 'trimData'; } if ($this->parser_options['XML_OPTION_CASE_FOLDING']==1) { $open_decor = new TSax3_CaseFolding( $this->handler_object_element, $this->handler_method_opening, $this->handler_method_closing); $this->handler_object_element =& $open_decor; $this->handler_method_opening ='foldOpen'; $this->handler_method_closing ='foldClose'; } if ($this->parser_options['XML_OPTION_LINEFEED_BREAK']==1) { $decorator = new TSax3_Linefeed( $this->handler_object_data, $this->handler_method_data); $this->handler_object_data =& $decorator; $this->handler_method_data = 'breakData'; } if ($this->parser_options['XML_OPTION_TAB_BREAK']==1) { $decorator = new TSax3_Tab( $this->handler_object_data, $this->handler_method_data); $this->handler_object_data =& $decorator; $this->handler_method_data = 'breakData'; } if ($this->parser_options['XML_OPTION_ENTITIES_UNPARSED']==1) { $decorator = new TSax3_Entities_Unparsed( $this->handler_object_data, $this->handler_method_data); $this->handler_object_data =& $decorator; $this->handler_method_data = 'breakData'; } if ($this->parser_options['XML_OPTION_ENTITIES_PARSED']==1) { $decorator = new TSax3_Entities_Parsed( $this->handler_object_data, $this->handler_method_data); $this->handler_object_data =& $decorator; $this->handler_method_data = 'breakData'; } // Note switched on by default if ($this->parser_options['XML_OPTION_STRIP_ESCAPES']==1) { $decorator = new TSax3_Escape_Stripper( $this->handler_object_escape, $this->handler_method_escape); $this->handler_object_escape =& $decorator; $this->handler_method_escape = 'strip'; } $this->rawtext = $data; $this->length = strlen($data); $this->position = 0; $this->_parse(); } /** * Performs the parsing itself, delegating calls to a specific parser * state * @param constant state object to parse with * @access protected * @return void */ function _parse($state = self::TSAX3_STATE_START) { do { $state = $this->State[$state]->parse($this); } while ($state != self::TSAX3_STATE_STOP && $this->position < $this->length); } } /** * Parser for PHP Versions below 4.3.0. Uses a slower parsing mechanism than * the equivalent PHP 4.3.0+ subclass of StateParser * @package System.Security.SafeHtml * @access protected * @see TSax3_StateParser_Gtet430 */ class TSax3_StateParser_Lt430 extends TSax3_StateParser { /** * Constructs TSax3_StateParser_Lt430 defining available * parser options * @var TSax3 instance of user front end class * @access protected */ function __construct(& $htmlsax) { parent::__construct($htmlsax); $this->parser_options['XML_OPTION_TRIM_DATA_NODES'] = 0; $this->parser_options['XML_OPTION_CASE_FOLDING'] = 0; $this->parser_options['XML_OPTION_LINEFEED_BREAK'] = 0; $this->parser_options['XML_OPTION_TAB_BREAK'] = 0; $this->parser_options['XML_OPTION_ENTITIES_PARSED'] = 0; $this->parser_options['XML_OPTION_ENTITIES_UNPARSED'] = 0; $this->parser_options['XML_OPTION_STRIP_ESCAPES'] = 0; //var_dump($this->parser_options); } /** * Returns a string from the current position until the first instance of * one of the characters in the supplied string argument * @param string string to search until * @access protected * @return string */ function scanUntilCharacters($string) { $startpos = $this->position; while ($this->position < $this->length && strpos($string, $this->rawtext{$this->position}) === FALSE) { $this->position++; } return substr($this->rawtext, $startpos, $this->position - $startpos); } /** * Moves the position forward past any whitespace characters * @access protected * @return void */ function ignoreWhitespace() { while ($this->position < $this->length && strpos(" \n\r\t", $this->rawtext{$this->position}) !== FALSE) { $this->position++; } } /** * Begins the parsing operation, setting up the unparsed XML entities * decorator if necessary then delegating further work to parent * @param string XML document to parse * @access protected * @return void */ function parse($data) { parent::parse($data); } } /** * Parser for PHP Versions equal to or greater than 4.3.0. Uses a faster * parsing mechanism than the equivalent PHP < 4.3.0 subclass of StateParser * @package System.Security.SafeHtml * @access protected * @see TSax3_StateParser_Lt430 */ class TSax3_StateParser_Gtet430 extends TSax3_StateParser { /** * Constructs TSax3_StateParser_Gtet430 defining available * parser options * @var TSax3 instance of user front end class * @access protected */ function __construct(& $htmlsax) { parent::__construct($htmlsax); $this->parser_options['XML_OPTION_TRIM_DATA_NODES'] = 0; $this->parser_options['XML_OPTION_CASE_FOLDING'] = 0; $this->parser_options['XML_OPTION_LINEFEED_BREAK'] = 0; $this->parser_options['XML_OPTION_TAB_BREAK'] = 0; $this->parser_options['XML_OPTION_ENTITIES_PARSED'] = 0; $this->parser_options['XML_OPTION_ENTITIES_UNPARSED'] = 0; $this->parser_options['XML_OPTION_STRIP_ESCAPES'] = 0; } /** * Returns a string from the current position until the first instance of * one of the characters in the supplied string argument. * @param string string to search until * @access protected * @return string */ function scanUntilCharacters($string) { $startpos = $this->position; $length = strcspn($this->rawtext, $string, $startpos); $this->position += $length; return substr($this->rawtext, $startpos, $length); } /** * Moves the position forward past any whitespace characters * @access protected * @return void */ function ignoreWhitespace() { $this->position += strspn($this->rawtext, " \n\r\t", $this->position); } /** * Begins the parsing operation, setting up the parsed and unparsed * XML entity decorators if necessary then delegating further work * to parent * @param string XML document to parse * @access protected * @return void */ function parse($data) { parent::parse($data); } } /** * Default NullHandler for methods which were not set by user * @package System.Security.SafeHtml * @access protected */ class TSax3_NullHandler { /** * Generic handler method which does nothing * @access protected * @return void */ function DoNothing() { } } /** * User interface class. All user calls should only be made to this class * @package System.Security.SafeHtml * @access public */ class TSax3 { /** * Instance of concrete subclass of TSax3_StateParser * @var TSax3_StateParser * @access private */ private $state_parser; /** * Constructs TSax3 selecting concrete StateParser subclass * depending on PHP version being used as well as setting the default * NullHandler for all callbacks
* Example: * 
    * $myHandler = & new MyHandler();
    * $parser = new TSax3();
    * $parser->set_object($myHandler);
    * $parser->set_option('XML_OPTION_CASE_FOLDING');
    * $parser->set_element_handler('myOpenHandler','myCloseHandler');
    * $parser->set_data_handler('myDataHandler');
    * $parser->parser($xml);
    *
* @access public */ function __construct() { if (version_compare(phpversion(), '4.3', 'ge')) { $this->state_parser = new TSax3_StateParser_Gtet430($this); } else { $this->state_parser = new TSax3_StateParser_Lt430($this); } $nullhandler = new TSax3_NullHandler(); $this->set_object($nullhandler); $this->set_element_handler('DoNothing', 'DoNothing'); $this->set_data_handler('DoNothing'); $this->set_pi_handler('DoNothing'); $this->set_jasp_handler('DoNothing'); $this->set_escape_handler('DoNothing'); } /** * Sets the user defined handler object. Returns a PEAR Error * if supplied argument is not an object. * @param object handler object containing SAX callback methods * @access public * @return mixed */ function set_object(&$object) { if ( is_object($object) ) { $this->state_parser->handler_default =& $object; return true; } else { require_once('PEAR.php'); PEAR::raiseError('TSax3::set_object requires '. 'an object instance'); } } /** * Sets a parser option. By default all options are switched off. * Returns a PEAR Error if option is invalid
* Available options: * * To get HTMLSax to behave in the same way as the native PHP SAX parser, * using it's default state, you need to switch on XML_OPTION_LINEFEED_BREAK, * XML_OPTION_ENTITIES_PARSED and XML_OPTION_CASE_FOLDING * @param string name of parser option * @param int (optional) 1 to switch on, 0 for off * @access public * @return boolean */ function set_option($name, $value=1) { if ( array_key_exists($name,$this->state_parser->parser_options) ) { $this->state_parser->parser_options[$name] = $value; return true; } else { require_once('PEAR.php'); PEAR::raiseError('TSax3::set_option('.$name.') illegal'); } } /** * Sets the data handler method which deals with the contents of XML * elements.
* The handler method must accept two arguments, the first being an * instance of TSax3 and the second being the contents of an * XML element e.g. * 
    * function myDataHander(& $parser,$data){}
    *
* @param string name of method * @access public * @return void * @see set_object */ function set_data_handler($data_method) { $this->state_parser->handler_object_data =& $this->state_parser->handler_default; $this->state_parser->handler_method_data = $data_method; } /** * Sets the open and close tag handlers *
The open handler method must accept three arguments; the parser, * the tag name and an array of attributes e.g. * 
    * function myOpenHander(& $parser,$tagname,$attrs=array()){}
    *
* The close handler method must accept two arguments; the parser and * the tag name e.g. * 
    * function myCloseHander(& $parser,$tagname){}
    *
* @param string name of open method * @param string name of close method * @access public * @return void * @see set_object */ function set_element_handler($opening_method, $closing_method) { $this->state_parser->handler_object_element =& $this->state_parser->handler_default; $this->state_parser->handler_method_opening = $opening_method; $this->state_parser->handler_method_closing = $closing_method; } /** * Sets the processing instruction handler method e.g. for PHP open * and close tags
* The handler method must accept three arguments; the parser, the * PI target and data inside the PI * 
    * function myPIHander(& $parser,$target, $data){}
    *
* @param string name of method * @access public * @return void * @see set_object */ function set_pi_handler($pi_method) { $this->state_parser->handler_object_pi =& $this->state_parser->handler_default; $this->state_parser->handler_method_pi = $pi_method; } /** * Sets the XML escape handler method e.g. for comments and doctype * declarations
* The handler method must accept two arguments; the parser and the * contents of the escaped section * 
    * function myEscapeHander(& $parser, $data){}
    *
* @param string name of method * @access public * @return void * @see set_object */ function set_escape_handler($escape_method) { $this->state_parser->handler_object_escape =& $this->state_parser->handler_default; $this->state_parser->handler_method_escape = $escape_method; } /** * Sets the JSP/ASP markup handler
* The handler method must accept two arguments; the parser and * body of the JASP tag * 
    * function myJaspHander(& $parser, $data){}
    *
* @param string name of method * @access public * @return void * @see set_object */ function set_jasp_handler ($jasp_method) { $this->state_parser->handler_object_jasp =& $this->state_parser->handler_default; $this->state_parser->handler_method_jasp = $jasp_method; } /** * Returns the current string position of the "cursor" inside the XML * document *
Intended for use from within a user defined handler called * via the $parser reference e.g. * 
    * function myDataHandler(& $parser,$data) {
    *     echo( 'Current position: '.$parser->get_current_position() );
    * }
    *
* @access public * @return int * @see get_length */ function get_current_position() { return $this->state_parser->position; } /** * Returns the string length of the XML document being parsed * @access public * @return int */ function get_length() { return $this->state_parser->length; } /** * Start parsing some XML * @param string XML document * @access public * @return void */ function parse($data) { $this->state_parser->parse($data); } }