[Sommaire] [Imprimer]

   1  <?php
   2  /**
   3  * SAXY_Parser_Base is a base class for SAXY and SAXY Lite
   4  * @package saxy-xmlparser
   5  * @version 1.0
   6  * @copyright (C) 2004 John Heinstein. All rights reserved
   7  * @license http://www.gnu.org/copyleft/lesser.html LGPL License
   8  * @author John Heinstein <johnkarl@nbnet.nb.ca>
   9  * @link http://www.engageinteractive.com/saxy/ SAXY Home Page
  10  * SAXY is Free Software
  11  **/
  12  
  13  /** the initial characters of a cdata section */
  14  define('SAXY_SEARCH_CDATA', '![CDATA[');
  15  /** the length of the initial characters of a cdata section */
  16  define('SAXY_CDATA_LEN', 8);
  17  /** the initial characters of a notation */
  18  define('SAXY_SEARCH_NOTATION', '!NOTATION');
  19  /** the initial characters of a doctype */
  20  define('SAXY_SEARCH_DOCTYPE', '!DOCTYPE');
  21  /** saxy parse state, just before parsing an attribute */
  22  define('SAXY_STATE_ATTR_NONE', 0);
  23  /** saxy parse state, parsing an attribute key */
  24  define('SAXY_STATE_ATTR_KEY', 1);
  25  /** saxy parse state, parsing an attribute value */
  26  define('SAXY_STATE_ATTR_VALUE', 2);
  27  
  28  /**
  29  * The base SAX Parser class
  30  *
  31  * @package saxy-xmlparser
  32  * @author John Heinstein <johnkarl@nbnet.nb.ca>
  33  */
  34  class SAXY_Parser_Base {
  35      /** @var int The current state of the parser */
  36      var $state;
  37      /** @var int A temporary container for parsed characters */
  38      var $charContainer;
  39      /** @var Object A reference to the start event handler */
  40      var $startElementHandler;
  41      /** @var Object A reference to the end event handler */
  42      var $endElementHandler;
  43      /** @var Object A reference to the data event handler */
  44      var $characterDataHandler;
  45      /** @var Object A reference to the CDATA Section event handler */
  46      var $cDataSectionHandler = null;
  47      /** @var boolean True if predefined entities are to be converted into characters */
  48      var $convertEntities = true;
  49      /** @var Array Translation table for predefined entities */
  50      var $predefinedEntities = array('&amp;' => '&', '&lt;' => '<', '&gt;' => '>',
  51                              '&quot;' => '"', '&apos;' => "'");
  52      /** @var Array User defined translation table for entities */
  53      var $definedEntities = array();
  54      /** @var boolean True if whitespace is to be preserved during parsing. NOT YET IMPLEMENTED! */
  55      var $preserveWhitespace = false;
  56  
  57  
  58      /**
  59      * Constructor for SAX parser
  60      */
  61  	function SAXY_Parser_Base() {
  62          $this->charContainer = '';
  63      } //SAXY_Parser_Base
  64  
  65      /**
  66      * Sets a reference to the handler for the start element event
  67      * @param mixed A reference to the start element handler
  68      */
  69  	function xml_set_element_handler($startHandler, $endHandler) {
  70          $this->startElementHandler = $startHandler;
  71          $this->endElementHandler = $endHandler;
  72      } //xml_set_element_handler
  73  
  74      /**
  75      * Sets a reference to the handler for the data event
  76      * @param mixed A reference to the data handler
  77      */
  78  	function xml_set_character_data_handler($handler) {
  79          $this->characterDataHandler =& $handler;
  80      } //xml_set_character_data_handler
  81  
  82      /**
  83      * Sets a reference to the handler for the CDATA Section event
  84      * @param mixed A reference to the CDATA Section handler
  85      */
  86  	function xml_set_cdata_section_handler($handler) {
  87          $this->cDataSectionHandler =& $handler;
  88      } //xml_set_cdata_section_handler
  89  
  90      /**
  91      * Sets whether predefined entites should be replaced with their equivalent characters during parsing
  92      * @param boolean True if entity replacement is to occur
  93      */
  94  	function convertEntities($truthVal) {
  95          $this->convertEntities = $truthVal;
  96      } //convertEntities
  97  
  98      /**
  99      * Appends an array of entity mappings to the existing translation table
 100      *
 101      * Intended mainly to facilitate the conversion of non-ASCII entities into equivalent characters
 102      *
 103      * @param array A list of entity mappings in the format: array('&amp;' => '&');
 104      */
 105  	function appendEntityTranslationTable($table) {
 106          $this->definedEntities = $table;
 107      } //appendEntityTranslationTable
 108  
 109  
 110      /**
 111      * Gets the nth character from the end of the string
 112      * @param string The text to be queried
 113      * @param int The index from the end of the string
 114      * @return string The found character
 115      */
 116  	function getCharFromEnd($text, $index) {
 117          $len = strlen($text);
 118          $char = $text{($len - 1 - $index)};
 119  
 120          return $char;
 121      } //getCharFromEnd
 122  
 123      /**
 124      * Parses the attributes string into an array of key / value pairs
 125      * @param string The attribute text
 126      * @return Array An array of key / value pairs
 127      */
 128  	function parseAttributes($attrText) {
 129          $attrText = trim($attrText);
 130          $attrArray = array();
 131          $maybeEntity = false;
 132  
 133          $total = strlen($attrText);
 134          $keyDump = '';
 135          $valueDump = '';
 136          $currentState = SAXY_STATE_ATTR_NONE;
 137          $quoteType = '';
 138  
 139          for ($i = 0; $i < $total; $i++) {
 140              $currentChar = $attrText{$i};
 141  
 142              if ($currentState == SAXY_STATE_ATTR_NONE) {
 143                  if (trim($currentChar != '')) {
 144                      $currentState = SAXY_STATE_ATTR_KEY;
 145                  }
 146              }
 147  
 148              switch ($currentChar) {
 149                  case "\t":
 150                      if ($currentState == SAXY_STATE_ATTR_VALUE) {
 151                          $valueDump .= $currentChar;
 152                      }
 153                      else {
 154                          $currentChar = '';
 155                      }
 156                      break;
 157  
 158                  case "\x0B": //vertical tab
 159                  case "\n":
 160                  case "\r":
 161                      $currentChar = '';
 162                      break;
 163  
 164                  case '=':
 165                      if ($currentState == SAXY_STATE_ATTR_VALUE) {
 166                          $valueDump .= $currentChar;
 167                      }
 168                      else {
 169                          $currentState = SAXY_STATE_ATTR_VALUE;
 170                          $quoteType = '';
 171                          $maybeEntity = false;
 172                      }
 173                      break;
 174  
 175                  case '"':
 176                      if ($currentState == SAXY_STATE_ATTR_VALUE) {
 177                          if ($quoteType == '') {
 178                              $quoteType = '"';
 179                          }
 180                          else {
 181                              if ($quoteType == $currentChar) {
 182                                  // Joomla! hack
 183                                  if (isset( $this ) && $this->convertEntities && $maybeEntity) {
 184                                      $valueDump = strtr($valueDump, $this->predefinedEntities);
 185                                      $valueDump = strtr($valueDump, $this->definedEntities);
 186                                  }
 187  
 188                                  $keyDump = trim($keyDump);
 189                                  $attrArray[$keyDump] = $valueDump;
 190                                  $keyDump = $valueDump = $quoteType = '';
 191                                  $currentState = SAXY_STATE_ATTR_NONE;
 192                              }
 193                              else {
 194                                  $valueDump .= $currentChar;
 195                              }
 196                          }
 197                      }
 198                      break;
 199  
 200                  case "'":
 201                      if ($currentState == SAXY_STATE_ATTR_VALUE) {
 202                          if ($quoteType == '') {
 203                              $quoteType = "'";
 204                          }
 205                          else {
 206                              if ($quoteType == $currentChar) {
 207                                  // Joomla! hack
 208                                  if (isset( $this ) && $this->convertEntities && $maybeEntity) {
 209                                      $valueDump = strtr($valueDump, $this->predefinedEntities);
 210                                      $valueDump = strtr($valueDump, $this->definedEntities);
 211                                  }
 212  
 213                                  $keyDump = trim($keyDump);
 214                                  $attrArray[$keyDump] = $valueDump;
 215                                  $keyDump = $valueDump = $quoteType = '';
 216                                  $currentState = SAXY_STATE_ATTR_NONE;
 217                              }
 218                              else {
 219                                  $valueDump .= $currentChar;
 220                              }
 221                          }
 222                      }
 223                      break;
 224  
 225                  case '&':
 226                      //might be an entity
 227                      $maybeEntity = true;
 228                      $valueDump .= $currentChar;
 229                      break;
 230  
 231                  default:
 232                      if ($currentState == SAXY_STATE_ATTR_KEY) {
 233                          $keyDump .= $currentChar;
 234                      }
 235                      else {
 236                          $valueDump .= $currentChar;
 237                      }
 238              }
 239          }
 240  
 241          return $attrArray;
 242      } //parseAttributes
 243  
 244      /**
 245      * Parses character data
 246      * @param string The character data
 247      */
 248  	function parseBetweenTags($betweenTagText) {
 249          if (trim($betweenTagText) != ''){
 250              $this->fireCharacterDataEvent($betweenTagText);
 251          }
 252      } //parseBetweenTags
 253  
 254      /**
 255      * Fires a start element event
 256      * @param string The start element tag name
 257      * @param Array The start element attributes
 258      */
 259  	function fireStartElementEvent($tagName, $attributes) {
 260          call_user_func($this->startElementHandler, $this, $tagName, $attributes);
 261      } //fireStartElementEvent
 262  
 263      /**
 264      * Fires an end element event
 265      * @param string The end element tag name
 266      */
 267  	function fireEndElementEvent($tagName) {
 268          call_user_func($this->endElementHandler, $this, $tagName);
 269      } //fireEndElementEvent
 270  
 271      /**
 272      * Fires a character data event
 273      * @param string The character data
 274      */
 275  	function fireCharacterDataEvent($data) {
 276          if ($this->convertEntities && ((strpos($data, "&") != -1))) {
 277              $data = strtr($data, $this->predefinedEntities);
 278              $data = strtr($data, $this->definedEntities);
 279          }
 280  
 281          call_user_func($this->characterDataHandler, $this, $data);
 282      } //fireCharacterDataEvent
 283  
 284      /**
 285      * Fires a CDATA Section event
 286      * @param string The CDATA Section data
 287      */
 288  	function fireCDataSectionEvent($data) {
 289          call_user_func($this->cDataSectionHandler, $this, $data);
 290      } //fireCDataSectionEvent
 291  } //SAXY_Parser_Base
 292  
 293  ?>