www/extensions/safehtml/lib/safehtml/classes/HTMLSax3.php

   1 <?php
   2 // +----------------------------------------------------------------------+
   3 // | Copyright (c) 1997-2002 The PHP Group                                |
   4 // +----------------------------------------------------------------------+
   5 // | This source file is subject to version 2.02 of the PHP license,      |
   6 // | that is bundled with this package in the file LICENSE, and is        |
   7 // | available at through the world-wide-web at                           |
   8 // | http://www.php.net/license/3_0.txt.                                  |
   9 // | If you did not receive a copy of the PHP license and are unable to   |
  10 // | obtain it through the world-wide-web, please send a note to          |
  11 // | license@php.net so we can mail you a copy immediately.               |
  12 // +----------------------------------------------------------------------+
  13 // | Authors: Alexander Zhukov <alex@veresk.ru> Original port from Python |
  14 // | Authors: Harry Fuecks <hfuecks@phppatterns.com> Port to PEAR + more  |
  15 // | Authors: Many @ Sitepointforums Advanced PHP Forums                  |
  16 // +----------------------------------------------------------------------+
  17 //
  18
  19 if (!defined('_ECRIRE_INC_VERSION')) return;
  20
  21 if (!defined('XML_HTMLSAX3')) {
  22  define('XML_HTMLSAX3', 'XML/');
  23 }
  24 require_once(XML_HTMLSAX3 . 'HTMLSax3/States.php');
  25 require_once(XML_HTMLSAX3 . 'HTMLSax3/Decorators.php');
  26
  27 class XML_HTMLSax3_StateParser {
  28  var $htmlsax;
  29  var $handler_object_element;
  30  var $handler_method_opening;
  31  var $handler_method_closing;
  32  var $handler_object_data;
  33  var $handler_method_data;
  34  var $handler_object_pi;
  35  var $handler_method_pi;
  36  var $handler_object_jasp;
  37  var $handler_method_jasp;
  38  var $handler_object_escape;
  39  var $handler_method_escape;
  40  var $handler_default;
  41  var $parser_options = array();
  42  var $rawtext;
  43  var $position;
  44  var $length;
  45  var $State = array();
  46
  47  function XML_HTMLSax3_StateParser (& $htmlsax) {
  48   $this->htmlsax = & $htmlsax;
  49
  50   $this->State[XML_HTMLSAX3_STATE_START] = new XML_HTMLSax3_StartingState();
  51
  52   $this->State[XML_HTMLSAX3_STATE_CLOSING_TAG] = new XML_HTMLSax3_ClosingTagState();
  53   $this->State[XML_HTMLSAX3_STATE_TAG] = new XML_HTMLSax3_TagState();
  54   $this->State[XML_HTMLSAX3_STATE_OPENING_TAG] = new XML_HTMLSax3_OpeningTagState();
  55
  56   $this->State[XML_HTMLSAX3_STATE_PI] = new XML_HTMLSax3_PiState();
  57   $this->State[XML_HTMLSAX3_STATE_JASP] = new XML_HTMLSax3_JaspState();
  58   $this->State[XML_HTMLSAX3_STATE_ESCAPE] = new XML_HTMLSax3_EscapeState();
  59  }
  60
  61  function unscanCharacter() {
  62   $this->position -= 1;
  63  }
  64
  65  function ignoreCharacter() {
  66   $this->position += 1;
  67  }
  68
  69  function scanCharacter() {
  70   if ($this->position < $this->length) {
  71    return $this->rawtext{$this->position++};
  72   }
  73  }
  74
  75  function scanUntilString($string) {
  76   $start = $this->position;
  77   $this->position = strpos($this->rawtext, $string, $start);
  78   if ($this->position === FALSE) {
  79    $this->position = $this->length;
  80   }
  81   return substr($this->rawtext, $start, $this->position - $start);
  82  }
  83
  84  function scanUntilCharacters($string) {}
  85
  86  function ignoreWhitespace() {}
  87
  88  function parse($data) {
  89   if ($this->parser_options['XML_OPTION_TRIM_DATA_NODES']==1) {
  90    $decorator = new XML_HTMLSax3_Trim(
  91     $this->handler_object_data,
  92     $this->handler_method_data);
  93    $this->handler_object_data =& $decorator;
  94    $this->handler_method_data = 'trimData';
  95   }
  96   if ($this->parser_options['XML_OPTION_CASE_FOLDING']==1) {
  97    $open_decor = new XML_HTMLSax3_CaseFolding(
  98     $this->handler_object_element,
  99     $this->handler_method_opening,
 100     $this->handler_method_closing);
 101    $this->handler_object_element =& $open_decor;
 102    $this->handler_method_opening ='foldOpen';
 103    $this->handler_method_closing ='foldClose';
 104   }
 105   if ($this->parser_options['XML_OPTION_LINEFEED_BREAK']==1) {
 106    $decorator = new XML_HTMLSax3_Linefeed(
 107     $this->handler_object_data,
 108     $this->handler_method_data);
 109    $this->handler_object_data =& $decorator;
 110    $this->handler_method_data = 'breakData';
 111   }
 112   if ($this->parser_options['XML_OPTION_TAB_BREAK']==1) {
 113    $decorator = new XML_HTMLSax3_Tab(
 114     $this->handler_object_data,
 115     $this->handler_method_data);
 116    $this->handler_object_data =& $decorator;
 117    $this->handler_method_data = 'breakData';
 118   }
 119   if ($this->parser_options['XML_OPTION_ENTITIES_UNPARSED']==1) {
 120    $decorator = new XML_HTMLSax3_Entities_Unparsed(
 121     $this->handler_object_data,
 122     $this->handler_method_data);
 123    $this->handler_object_data =& $decorator;
 124    $this->handler_method_data = 'breakData';
 125   }
 126   if ($this->parser_options['XML_OPTION_ENTITIES_PARSED']==1) {
 127    $decorator = new XML_HTMLSax3_Entities_Parsed(
 128     $this->handler_object_data,
 129     $this->handler_method_data);
 130    $this->handler_object_data =& $decorator;
 131    $this->handler_method_data = 'breakData';
 132   }
 133   // Note switched on by default
 134   if ($this->parser_options['XML_OPTION_STRIP_ESCAPES']==1) {
 135    $decorator = new XML_HTMLSax3_Escape_Stripper(
 136     $this->handler_object_escape,
 137     $this->handler_method_escape);
 138    $this->handler_object_escape =& $decorator;
 139    $this->handler_method_escape = 'strip';
 140   }
 141   $this->rawtext = $data;
 142   $this->length = strlen($data);
 143   $this->position = 0;
 144   $this->_parse();
 145  }
 146
 147  function _parse($state = XML_HTMLSAX3_STATE_START) {
 148   do {
 149    $state = $this->State[$state]->parse($this);
 150   } while ($state != XML_HTMLSAX3_STATE_STOP &&
 151      $this->position < $this->length);
 152  }
 153 }
 154
 155 class XML_HTMLSax3_StateParser_Lt430 extends XML_HTMLSax3_StateParser {
 156  function XML_HTMLSax3_StateParser_Lt430(& $htmlsax) {
 157   parent::XML_HTMLSax3_StateParser($htmlsax);
 158   $this->parser_options['XML_OPTION_TRIM_DATA_NODES'] = 0;
 159   $this->parser_options['XML_OPTION_CASE_FOLDING'] = 0;
 160   $this->parser_options['XML_OPTION_LINEFEED_BREAK'] = 0;
 161   $this->parser_options['XML_OPTION_TAB_BREAK'] = 0;
 162   $this->parser_options['XML_OPTION_ENTITIES_PARSED'] = 0;
 163   $this->parser_options['XML_OPTION_ENTITIES_UNPARSED'] = 0;
 164   $this->parser_options['XML_OPTION_STRIP_ESCAPES'] = 0;
 165  }
 166
 167  function scanUntilCharacters($string) {
 168   $startpos = $this->position;
 169   while ($this->position < $this->length && strpos($string, $this->rawtext{$this->position}) === FALSE) {
 170    $this->position++;
 171   }
 172   return substr($this->rawtext, $startpos, $this->position - $startpos);
 173  }
 174
 175  function ignoreWhitespace() {
 176   while ($this->position < $this->length &&
 177    strpos(" \n\r\t", $this->rawtext{$this->position}) !== FALSE) {
 178    $this->position++;
 179   }
 180  }
 181
 182  function parse($data) {
 183   parent::parse($data);
 184  }
 185 }
 186
 187 class XML_HTMLSax3_StateParser_Gtet430 extends XML_HTMLSax3_StateParser {
 188  function XML_HTMLSax3_StateParser_Gtet430(& $htmlsax) {
 189   parent::XML_HTMLSax3_StateParser($htmlsax);
 190   $this->parser_options['XML_OPTION_TRIM_DATA_NODES'] = 0;
 191   $this->parser_options['XML_OPTION_CASE_FOLDING'] = 0;
 192   $this->parser_options['XML_OPTION_LINEFEED_BREAK'] = 0;
 193   $this->parser_options['XML_OPTION_TAB_BREAK'] = 0;
 194   $this->parser_options['XML_OPTION_ENTITIES_PARSED'] = 0;
 195   $this->parser_options['XML_OPTION_ENTITIES_UNPARSED'] = 0;
 196   $this->parser_options['XML_OPTION_STRIP_ESCAPES'] = 0;
 197  }
 198  function scanUntilCharacters($string) {
 199   $startpos = $this->position;
 200   $length = strcspn($this->rawtext, $string, $startpos);
 201   $this->position += $length;
 202   return substr($this->rawtext, $startpos, $length);
 203  }
 204
 205  function ignoreWhitespace() {
 206   $this->position += strspn($this->rawtext, " \n\r\t", $this->position);
 207  }
 208
 209  function parse($data) {
 210   parent::parse($data);
 211  }
 212 }
 213
 214 class XML_HTMLSax3_NullHandler {
 215  function DoNothing() {
 216  }
 217 }
 218
 219 class XML_HTMLSax3 {
 220  var $state_parser;
 221
 222  function XML_HTMLSax3() {
 223   if (version_compare(phpversion(), '4.3', 'ge')) {
 224    $this->state_parser = new XML_HTMLSax3_StateParser_Gtet430($this);
 225   } else {
 226    $this->state_parser = new XML_HTMLSax3_StateParser_Lt430($this);
 227   }
 228   $nullhandler = new XML_HTMLSax3_NullHandler();
 229   $this->set_object($nullhandler);
 230   $this->set_element_handler('DoNothing', 'DoNothing');
 231   $this->set_data_handler('DoNothing');
 232   $this->set_pi_handler('DoNothing');
 233   $this->set_jasp_handler('DoNothing');
 234   $this->set_escape_handler('DoNothing');
 235  }
 236
 237  function set_object(&$object) {
 238   if ( is_object($object) ) {
 239    $this->state_parser->handler_default =& $object;
 240    return true;
 241   } else {
 242    require_once('PEAR.php');
 243    PEAR::raiseError('XML_HTMLSax3::set_object requires '.
 244     'an object instance');
 245   }
 246  }
 247
 248  function set_option($name, $value=1) {
 249   if ( array_key_exists($name,$this->state_parser->parser_options) ) {
 250    $this->state_parser->parser_options[$name] = $value;
 251    return true;
 252   } else {
 253    require_once('PEAR.php');
 254    PEAR::raiseError('XML_HTMLSax3::set_option('.$name.') illegal');
 255   }
 256  }
 257
 258  function set_data_handler($data_method) {
 259   $this->state_parser->handler_object_data =& $this->state_parser->handler_default;
 260   $this->state_parser->handler_method_data = $data_method;
 261  }
 262
 263  function set_element_handler($opening_method, $closing_method) {
 264   $this->state_parser->handler_object_element =& $this->state_parser->handler_default;
 265   $this->state_parser->handler_method_opening = $opening_method;
 266   $this->state_parser->handler_method_closing = $closing_method;
 267  }
 268
 269  function set_pi_handler($pi_method) {
 270   $this->state_parser->handler_object_pi =& $this->state_parser->handler_default;
 271   $this->state_parser->handler_method_pi = $pi_method;
 272  }
 273
 274  function set_escape_handler($escape_method) {
 275   $this->state_parser->handler_object_escape =& $this->state_parser->handler_default;
 276   $this->state_parser->handler_method_escape = $escape_method;
 277  }
 278
 279  function set_jasp_handler ($jasp_method) {
 280   $this->state_parser->handler_object_jasp =& $this->state_parser->handler_default;
 281   $this->state_parser->handler_method_jasp = $jasp_method;
 282  }
 283
 284  function get_current_position() {
 285   return $this->state_parser->position;
 286  }
 287
 288  function get_length() {
 289   return $this->state_parser->length;
 290  }
 291
 292  function parse($data) {
 293   $this->state_parser->parse($data);
 294  }
 295 }
 296 ?>