xml.php

   1 <?php
   2 /*=====================================================================*\
   3 || ################################################################### ||
   4 || # Iris Studios Shared Object Framework [#]version[#]
   5 || # --------------------------------------------------------------- # ||
   6 || # All parts of this file are ©2003-[#]year[#] Iris Studios, Inc. No     # ||
   7 || # part of this file may be reproduced in any way: part or whole.  # ||
   8 || # --------------------------------------------------------------- # ||
   9 || # ©2003 - [#]year[#] Iris Studios, Inc. | http://www.iris-studios.com   # ||
  10 || ################################################################### ||
  11 \*=====================================================================*/
  12
  13 $OBJECT = 'XML Parser';
  14 $CLASS = 'XML_Parser';
  15 $OBJ = 'xml';
  16
  17 /**
  18 * XML Parser
  19 *
  20 * This framework is a wrapper for a robust XML parser.
  21 *
  22 * @author               Iris Studios, Inc.
  23 * @copyright    Copyright ©2003 - [#]year[#], Iris Studios, Inc.
  24 * @version              $Revision$
  25 *
  26 */
  27 class XML_Parser
  28 {
  29         /**
  30         * Global environment variables
  31         *
  32         * @var  parser          Current XML parser
  33         * @var  taginfo         Array of all the processed tag data
  34         * @var  taghandler      What to do for each tag encountered
  35         * @var  trimdata        Trim CDATA elements?
  36         * @var  attrdata        Data parsed from global attributes
  37         * @var  tagid           Current tag ID in the document
  38         * @var  tagname         Current tag name in the document
  39         * @var  tree            Tag stack to use for proper nesting
  40         * @var  parent          The tag's parent
  41         */
  42         var $parser = null;
  43         var $taginfo = array();
  44         var $taghandler = array();
  45         var $trimdata = true;
  46         var $attrdata = array();
  47         var $tagid = -1;
  48         var $tagname = '';
  49         var $tree = array();
  50         var $parent = -1;
  51
  52         /**
  53         * Parse an XML file
  54         *
  55         * @param        str             XML file data
  56         *
  57         * @return       array   Array with all the XML data parsed
  58         */
  59         function parse($data)
  60         {
  61                 $this->parser = xml_parser_create();
  62                 xml_set_object($this->parser, $this);
  63                 xml_parser_set_option($this->parser, XML_OPTION_CASE_FOLDING, 0);
  64                 xml_set_element_handler($this->parser, '_tag_start', '_tag_end');
  65                 xml_set_character_data_handler($this->parser, '_cdata');
  66
  67                 if (!xml_parse($this->parser, $data))
  68                 {
  69                         $error['code'] = xml_get_error_code($this->parser);
  70                         $error['string'] = xml_error_string($error['code']);
  71                         $error['line'] = xml_get_current_line_number($this->parser);
  72                         $error['column'] = xml_get_current_column_number($this->parser);
  73                         print(sprintf("XML Error: %s (%d) at line %d colunn %d", $error['string'], $error['code'], $error['line'], $error['column']));
  74                         exit;
  75                 }
  76
  77                 xml_parser_free($this->parser);
  78
  79                 return array();
  80         }
  81
  82         /**
  83         * Process the opening location of an XML tag
  84         *
  85         * @param        res             XML parser
  86         * @param        str             Tag name
  87         * @param        array   Tag attributes
  88         */
  89         function _tag_start($parser, $name, $attributes)
  90         {
  91                 // we're opening a new tag
  92                 $this->tagid++;
  93                 $this->tagname = $name;
  94
  95                 // copy all tag attributes
  96                 array_walk($attributes, 'trim');
  97                 $this->attrdata[ $this->tagid ] = $attributes;
  98
  99                 $this->parent = current($this->tree);
 100                 if ($this->parent === false)
 101                 {
 102                         $this->parent = -1;
 103                 }
 104                 // advance the parent counter because the tree is yet to be updated
 105                 else
 106                 {
 107                         $this->parent++;
 108                 }
 109
 110                 // initialize the data set
 111                 $this->taginfo["$name"][ $this->parent ][ $this->tagid ] = '';
 112
 113                 // add the current tag into the tree
 114                 $this->tree[] = $this->tagid;
 115         }
 116
 117         /**
 118         * Process XML CDATA
 119         *
 120         * @param        res             XML parser
 121         * @param        str             CDATA from tag
 122         */
 123         function _cdata($parser, $data)
 124         {
 125                 global $_isso;
 126                 static $count;
 127
 128                 /*if (preg_replace('#(^[[:space:]]+|[[:space:]]+$)#', '', trim($data)) == '')
 129                 {
 130                         $count++;
 131                         $datanew = str_replace(array("\t", "\n", "\r", " "), array("{t}", "{n}", "{r}", "{s}"), $data);
 132                         $_isso->debug("strangechars[" . $this->tagid . "] = '$datanew'");
 133                         $data = $datanew . "||$count||";
 134                         //return;
 135                 }*/
 136
 137                 $_isso->debug("cdata[] = '$data'");
 138                 // read in the CDATA
 139                 if ($this->parent == $this->tagid)
 140                 {
 141                         if (!isset($this->taginfo[ $this->tagname ][ $this->parent - 1 ][ $this->tagid ]))
 142                         {
 143                                 $this->taginfo[ $this->tagname ][ $this->parent - 1 ][ $this->tagid ] = '';
 144                         }
 145
 146                         $this->taginfo[ $this->tagname ][ $this->parent - 1 ][ $this->tagid ] .= $data;
 147                 }
 148                 else
 149                 {
 150                         $this->taginfo[ $this->tagname ][ $this->parent ][ $this->tagid ] .= $data;
 151                 }
 152         }
 153
 154         /**
 155         * Process the closing of an XML tag
 156         *
 157         * @param        res             XML parser
 158         * @param        str             Tag name
 159         */
 160         function _tag_end($parser, $name)
 161         {
 162                 global $_isso;
 163
 164                 // fetch the data
 165                 if (isset($this->taginfo[ $this->tagname ][ $this->parent ][ $this->tagid ]))
 166                 {
 167                         $data = $this->taginfo[ $this->tagname ][ $this->parent ][ $this->tagid ];
 168                 }
 169                 else
 170                 {
 171                         return;
 172                 }
 173
 174                 if (trim($data) == '')
 175                 {
 176                         $_isso->debug("not going to bother with '" . $this->tagid . "'");
 177                 }
 178
 179                 // if we have a data handler, operate it now
 180                 if (isset($this->taghandler[ $this->tagname ]))
 181                 {
 182                         $_isso->debug("handler: {$this->taghandler[ " . $this->tagname . " ]}");
 183                         if (function_exists($this->taghandler[ $this->tagname ]))
 184                         {
 185                                 $data = $this->taghandler[ $this->tagname ]($data);
 186                         }
 187                 }
 188
 189                 // check for necessary trims
 190                 if ($this->trimdata)
 191                 {
 192                         // trim() doesn't take off the edge for more than a few white characters, preg_replace does; don't use [:space:] either...
 193                         //$data = preg_replace('#(^\s+|\s+$)#', '', $data);
 194                         //$data = preg_replace('#(^[[:space:]]+|[[:space:]]+$)#', '', trim($data));
 195                         $data = preg_replace('#^\s+#', '', $data);
 196                         $data = preg_replace('#\s+$#', '', $data);
 197                         $_isso->debug("trimming data ('" . $this->tagid . "'): '" . $data . "'");
 198                         //$data = trim($data);
 199                 }
 200
 201                 // don't want blank fields
 202                 if (trim($data) != '')
 203                 {
 204                         $this->taginfo[ $this->tagname ][ $this->parent ][ $this->tagid ] = $data;
 205                 }
 206                 else
 207                 {
 208                         unset($this->taginfo[ $this->tagname ][ $this->parent ][ $this->tagid ]);
 209                 }
 210
 211                 $_isso->debug("tree[] = " . current($this->tree));
 212
 213                 // make sure that we're always at the end of the tree
 214                 end($this->tree);
 215         }
 216 }
 217
 218 /*=====================================================================*\
 219 || ###################################################################
 220 || # $HeadURL$
 221 || # $Id$
 222 || ###################################################################
 223 \*=====================================================================*/
 224 ?>