vendor/domit/xml_saxy_lite_parser.php

   1 <?php
   2 /**
   3 * SAXY Lite is a non-validating, but lightweight and fast SAX parser for PHP, modelled on the Expat parser
   4 * @package saxy-xmlparser
   5 * @subpackage saxy-xmlparser-lite
   6 * @version 0.17
   7 * @copyright (C) 2004 John Heinstein. All rights reserved
   8 * @license http://www.gnu.org/copyleft/lesser.html LGPL License
   9 * @author John Heinstein <johnkarl@nbnet.nb.ca>
  10 * @link http://www.engageinteractive.com/saxy/ SAXY Home Page
  11 * SAXY is Free Software
  12 **/
  13
  14 if (!defined('SAXY_INCLUDE_PATH')) {
  15         define('SAXY_INCLUDE_PATH', (dirname(__FILE__) . "/"));
  16 }
  17
  18 /** current version of SAXY Lite */
  19 define ('SAXY_LITE_VERSION', '0.17');
  20
  21 /** initial saxy lite parse state, before anything is encountered */
  22 define('SAXY_STATE_NONE', 0);
  23 /** saxy lite parse state, processing main document */
  24 define('SAXY_STATE_PARSING', 1);
  25
  26 require_once(SAXY_INCLUDE_PATH . 'xml_saxy_shared.php');
  27
  28 /**
  29 * The SAX Parser class
  30 *
  31 * @package saxy-xmlparser
  32 * @subpackage saxy-xmlparser-lite
  33 * @author John Heinstein <johnkarl@nbnet.nb.ca>
  34 */
  35 class SAXY_Lite_Parser extends SAXY_Parser_Base {
  36         /**
  37         * Constructor for SAX parser
  38         */
  39         function SAXY_Lite_Parser() {
  40                 $this->SAXY_Parser_Base();
  41                 $this->state = SAXY_STATE_NONE;
  42         } //SAXY_Lite_Parser
  43
  44         /**
  45         * Returns the current version of SAXY Lite
  46         * @return Object The current version of SAXY Lite
  47         */
  48         function getVersion() {
  49                 return SAXY_LITE_VERSION;
  50         } //getVersion
  51
  52         /**
  53         * Processes the xml prolog, doctype, and any other nodes that exist outside of the main xml document
  54         * @param string The xml text to be processed
  55         * @return string The preprocessed xml text
  56         */
  57         function preprocessXML($xmlText) {
  58                 //strip prolog
  59                 $xmlText = trim($xmlText);
  60                 $total = strlen($xmlText);
  61
  62                 for ($i = 0; $i < $total; $i++) {
  63                         if ($xmlText{$i} == '<') {
  64                                 switch ($xmlText{($i + 1)}) {
  65                                         case '?':
  66                                         case '!':
  67                                                 break;
  68                                         default:
  69                                                 $this->state = SAXY_STATE_PARSING;
  70                                                 return (substr($xmlText, $i));
  71                                 }
  72                         }
  73                 }
  74         } //preprocessXML
  75
  76         /**
  77         * The controlling method for the parsing process
  78         * @param string The xml text to be processed
  79         * @return boolean True if parsing is successful
  80         */
  81         function parse ($xmlText) {
  82                 $xmlText = $this->preprocessXML($xmlText);
  83                 $total = strlen($xmlText);
  84
  85                 for ($i = 0; $i < $total; $i++) {
  86                         $currentChar = $xmlText{$i};
  87
  88                         switch ($this->state) {
  89                                 case SAXY_STATE_PARSING:
  90
  91                                         switch ($currentChar) {
  92                                                 case '<':
  93                                                         if (substr($this->charContainer, 0, SAXY_CDATA_LEN) == SAXY_SEARCH_CDATA) {
  94                                                                 $this->charContainer .= $currentChar;
  95                                                         }
  96                                                         else {
  97                                                                 $this->parseBetweenTags($this->charContainer);
  98                                                                 $this->charContainer = '';
  99                                                         }
 100                                                         break;
 101
 102                                                 case '>':
 103                                                         if ((substr($this->charContainer, 0, SAXY_CDATA_LEN) == SAXY_SEARCH_CDATA) &&
 104                                                                 !(($this->getCharFromEnd($this->charContainer, 0) == ']') &&
 105                                                                 ($this->getCharFromEnd($this->charContainer, 1) == ']'))) {
 106                                                                 $this->charContainer .= $currentChar;
 107                                                         }
 108                                                         else {
 109                                                                 $this->parseTag($this->charContainer);
 110                                                                 $this->charContainer = '';
 111                                                         }
 112                                                         break;
 113
 114                                                 default:
 115                                                         $this->charContainer .= $currentChar;
 116                                         }
 117
 118                                         break;
 119                         }
 120                 }
 121
 122                 return true;
 123         } //parse
 124
 125         /**
 126         * Parses an element tag
 127         * @param string The interior text of the element tag
 128         */
 129         function parseTag($tagText) {
 130                 $tagText = trim($tagText);
 131                 $firstChar = $tagText{0};
 132                 $myAttributes = array();
 133
 134                 switch ($firstChar) {
 135                         case '/':
 136                                 $tagName = substr($tagText, 1);
 137                                 $this->fireEndElementEvent($tagName);
 138                                 break;
 139
 140                         case '!':
 141                                 $upperCaseTagText = strtoupper($tagText);
 142
 143                                 if (strpos($upperCaseTagText, SAXY_SEARCH_CDATA) !== false) { //CDATA Section
 144                                         $total = strlen($tagText);
 145                                         $openBraceCount = 0;
 146                                         $textNodeText = '';
 147
 148                                         for ($i = 0; $i < $total; $i++) {
 149                                                 $currentChar = $tagText{$i};
 150
 151                                                 if (($currentChar == ']') && ($tagText{($i + 1)} == ']')) {
 152                                                         break;
 153                                                 }
 154                                                 else if ($openBraceCount > 1) {
 155                                                         $textNodeText .= $currentChar;
 156                                                 }
 157                                                 else if ($currentChar == '[') { //this won't be reached after the first open brace is found
 158                                                         $openBraceCount ++;
 159                                                 }
 160                                         }
 161
 162                                         if ($this->cDataSectionHandler == null) {
 163                                                 $this->fireCharacterDataEvent($textNodeText);
 164                                         }
 165                                         else {
 166                                                 $this->fireCDataSectionEvent($textNodeText);
 167                                         }
 168                                 }
 169                                 else if (strpos($upperCaseTagText, SAXY_SEARCH_NOTATION) !== false) { //NOTATION node, discard
 170                                         return;
 171                                 }
 172                                 else if (substr($tagText, 0, 2) == '!-') { //comment node, discard
 173                                         return;
 174                                 }
 175
 176                                 break;
 177
 178                         case '?':
 179                                 //Processing Instruction node, discard
 180                                 return;
 181
 182                         default:
 183                                 if ((strpos($tagText, '"') !== false) || (strpos($tagText, "'") !== false)) {
 184                                         $total = strlen($tagText);
 185                                         $tagName = '';
 186
 187                                         for ($i = 0; $i < $total; $i++) {
 188                                                 $currentChar = $tagText{$i};
 189
 190                                                 if (($currentChar == ' ') || ($currentChar == "\t") ||
 191                                                         ($currentChar == "\n") || ($currentChar == "\r") ||
 192                                                         ($currentChar == "\x0B")) {
 193                                                         $myAttributes = $this->parseAttributes(substr($tagText, $i));
 194                                                         break;
 195                                                 }
 196                                                 else {
 197                                                         $tagName .= $currentChar;
 198                                                 }
 199                                         }
 200
 201                                         if (strrpos($tagText, '/') == (strlen($tagText) - 1)) { //check $tagText, but send $tagName
 202                                                 $this->fireStartElementEvent($tagName, $myAttributes);
 203                                                 $this->fireEndElementEvent($tagName);
 204                                         }
 205                                         else {
 206                                                 $this->fireStartElementEvent($tagName, $myAttributes);
 207                                         }
 208                                 }
 209                                 else {
 210                                         if (strpos($tagText, '/') !== false) {
 211                                                 $tagText = trim(substr($tagText, 0, (strrchr($tagText, '/') - 1)));
 212                                                 $this->fireStartElementEvent($tagText, $myAttributes);
 213                                                 $this->fireEndElementEvent($tagText);
 214                                         }
 215                                         else {
 216                                                 $this->fireStartElementEvent($tagText, $myAttributes);
 217                                         }
 218                                 }
 219                 }
 220         } //parseTag
 221
 222         /**
 223         * Returns the current error code (non-functional for SAXY Lite)
 224         * @return int The current error code
 225         */
 226         function xml_get_error_code() {
 227                 return -1;
 228         } //xml_get_error_code
 229
 230         /**
 231         * Returns a textual description of the error code (non-functional for SAXY Lite)
 232         * @param int The error code
 233         * @return string The error message
 234         */
 235         function xml_error_string($code) {
 236                 return "";
 237         } //xml_error_string
 238 } //SAXY_Lite_Parser
 239 ?>