vendor/domit/xml_saxy_shared.php

   1 <?php
   2 /**
   3 * SAXY_Parser_Base is a base class for SAXY and SAXY Lite
   4 * @package saxy-xmlparser
   5 * @version 0.87
   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
  55
  56         /**
  57         * Constructor for SAX parser
  58         */
  59         function SAXY_Parser_Base() {
  60                 $this->charContainer = '';
  61         } //SAXY_Parser_Base
  62
  63         /**
  64         * Sets a reference to the handler for the start element event
  65         * @param mixed A reference to the start element handler
  66         */
  67         function xml_set_element_handler($startHandler, $endHandler) {
  68                 $this->startElementHandler = $startHandler;
  69                 $this->endElementHandler = $endHandler;
  70         } //xml_set_element_handler
  71
  72         /**
  73         * Sets a reference to the handler for the data event
  74         * @param mixed A reference to the data handler
  75         */
  76         function xml_set_character_data_handler($handler) {
  77                 $this->characterDataHandler =& $handler;
  78         } //xml_set_character_data_handler
  79
  80         /**
  81         * Sets a reference to the handler for the CDATA Section event
  82         * @param mixed A reference to the CDATA Section handler
  83         */
  84         function xml_set_cdata_section_handler($handler) {
  85                 $this->cDataSectionHandler =& $handler;
  86         } //xml_set_cdata_section_handler
  87
  88         /**
  89         * Sets whether predefined entites should be replaced with their equivalent characters during parsing
  90         * @param boolean True if entity replacement is to occur
  91         */
  92         function convertEntities($truthVal) {
  93                 $this->convertEntities = $truthVal;
  94         } //convertEntities
  95
  96         /**
  97         * Appends an array of entity mappings to the existing translation table
  98         *
  99         * Intended mainly to facilitate the conversion of non-ASCII entities into equivalent characters
 100         *
 101         * @param array A list of entity mappings in the format: array('&amp;' => '&');
 102         */
 103         function appendEntityTranslationTable($table) {
 104                 $this->definedEntities = $table;
 105         } //appendEntityTranslationTable
 106
 107
 108         /**
 109         * Gets the nth character from the end of the string
 110         * @param string The text to be queried
 111         * @param int The index from the end of the string
 112         * @return string The found character
 113         */
 114         function getCharFromEnd($text, $index) {
 115                 $len = strlen($text);
 116                 $char = $text{($len - 1 - $index)};
 117
 118                 return $char;
 119         } //getCharFromEnd
 120
 121         /**
 122         * Parses the attributes string into an array of key / value pairs
 123         * @param string The attribute text
 124         * @return Array An array of key / value pairs
 125         */
 126         function parseAttributes($attrText) {
 127                 $attrText = trim($attrText);
 128                 $attrArray = array();
 129                 $maybeEntity = false;
 130
 131                 $total = strlen($attrText);
 132                 $keyDump = '';
 133                 $valueDump = '';
 134                 $currentState = SAXY_STATE_ATTR_NONE;
 135                 $quoteType = '';
 136
 137                 for ($i = 0; $i < $total; $i++) {
 138                         $currentChar = $attrText{$i};
 139
 140                         if ($currentState == SAXY_STATE_ATTR_NONE) {
 141                                 if (trim($currentChar != '')) {
 142                                         $currentState = SAXY_STATE_ATTR_KEY;
 143                                 }
 144                         }
 145
 146                         switch ($currentChar) {
 147                                 case "\t":
 148                                         if ($currentState == SAXY_STATE_ATTR_VALUE) {
 149                                                 $valueDump .= $currentChar;
 150                                         }
 151                                         else {
 152                                                 $currentChar = '';
 153                                         }
 154                                         break;
 155
 156                                 case "\x0B": //vertical tab
 157                                 case "\n":
 158                                 case "\r":
 159                                         $currentChar = '';
 160                                         break;
 161
 162                                 case '=':
 163                                         if ($currentState == SAXY_STATE_ATTR_VALUE) {
 164                                                 $valueDump .= $currentChar;
 165                                         }
 166                                         else {
 167                                                 $currentState = SAXY_STATE_ATTR_VALUE;
 168                                                 $quoteType = '';
 169                                                 $maybeEntity = false;
 170                                         }
 171                                         break;
 172
 173                                 case '"':
 174                                         if ($currentState == SAXY_STATE_ATTR_VALUE) {
 175                                                 if ($quoteType == '') {
 176                                                         $quoteType = '"';
 177                                                 }
 178                                                 else {
 179                                                         if ($quoteType == $currentChar) {
 180                                                                 if ($this->convertEntities && $maybeEntity) {
 181                                                                     $valueDump = strtr($valueDump, $this->predefinedEntities);
 182                                                                         $valueDump = strtr($valueDump, $this->definedEntities);
 183                                                                 }
 184
 185                                                                 $attrArray[trim($keyDump)] = $valueDump;
 186                                                                 $keyDump = $valueDump = $quoteType = '';
 187                                                                 $currentState = SAXY_STATE_ATTR_NONE;
 188                                                         }
 189                                                         else {
 190                                                                 $valueDump .= $currentChar;
 191                                                         }
 192                                                 }
 193                                         }
 194                                         break;
 195
 196                                 case "'":
 197                                         if ($currentState == SAXY_STATE_ATTR_VALUE) {
 198                                                 if ($quoteType == '') {
 199                                                         $quoteType = "'";
 200                                                 }
 201                                                 else {
 202                                                         if ($quoteType == $currentChar) {
 203                                                                 if ($this->convertEntities && $maybeEntity) {
 204                                                                     $valueDump = strtr($valueDump, $this->predefinedEntities);
 205                                                                         $valueDump = strtr($valueDump, $this->definedEntities);
 206                                                                 }
 207
 208                                                                 $attrArray[trim($keyDump)] = $valueDump;
 209                                                                 $keyDump = $valueDump = $quoteType = '';
 210                                                                 $currentState = SAXY_STATE_ATTR_NONE;
 211                                                         }
 212                                                         else {
 213                                                                 $valueDump .= $currentChar;
 214                                                         }
 215                                                 }
 216                                         }
 217                                         break;
 218
 219                                 case '&':
 220                                         //might be an entity
 221                                         $maybeEntity = true;
 222                                         $valueDump .= $currentChar;
 223                                         break;
 224
 225                                 default:
 226                                         if ($currentState == SAXY_STATE_ATTR_KEY) {
 227                                                 $keyDump .= $currentChar;
 228                                         }
 229                                         else {
 230                                                 $valueDump .= $currentChar;
 231                                         }
 232                         }
 233                 }
 234
 235                 return $attrArray;
 236         } //parseAttributes
 237
 238         /**
 239         * Parses character data
 240         * @param string The character data
 241         */
 242         function parseBetweenTags($betweenTagText) {
 243                 if (trim($betweenTagText) != ''){
 244                         $this->fireCharacterDataEvent($betweenTagText);
 245                 }
 246         } //parseBetweenTags
 247
 248         /**
 249         * Fires a start element event
 250         * @param string The start element tag name
 251         * @param Array The start element attributes
 252         */
 253         function fireStartElementEvent($tagName, $attributes) {
 254                 call_user_func($this->startElementHandler, $this, $tagName, $attributes);
 255         } //fireStartElementEvent
 256
 257         /**
 258         * Fires an end element event
 259         * @param string The end element tag name
 260         */
 261         function fireEndElementEvent($tagName) {
 262                 call_user_func($this->endElementHandler, $this, $tagName);
 263         } //fireEndElementEvent
 264
 265         /**
 266         * Fires a character data event
 267         * @param string The character data
 268         */
 269         function fireCharacterDataEvent($data) {
 270                 if ($this->convertEntities && ((strpos($data, "&") != -1))) {
 271                         $data = strtr($data, $this->predefinedEntities);
 272                         $data = strtr($data, $this->definedEntities);
 273                 }
 274
 275                 call_user_func($this->characterDataHandler, $this, $data);
 276         } //fireCharacterDataEvent
 277
 278         /**
 279         * Fires a CDATA Section event
 280         * @param string The CDATA Section data
 281         */
 282         function fireCDataSectionEvent($data) {
 283                 call_user_func($this->cDataSectionHandler, $this, $data);
 284         } //fireCDataSectionEvent
 285 } //SAXY_Parser_Base
 286
 287 ?>