1: <?php
  2: /**
  3:  * Handles parsing the provided XML input.
  4:  *
  5:  * PHP version 5
  6:  *
  7:  * @category Kolab
  8:  * @package  Kolab_Format
  9:  * @author   Gunnar Wrobel <wrobel@pardus.de>
 10:  * @license  http://www.horde.org/licenses/lgpl21 LGPL 2.1
 11:  * @link     http://www.horde.org/libraries/Horde_Kolab_Format
 12:  */
 13: 
 14: /**
 15:  * Handles parsing the provided XML input.
 16:  *
 17:  * Copyright 2007-2009 Klarälvdalens Datakonsult AB
 18:  * Copyright 2010-2012 Horde LLC (http://www.horde.org/)
 19:  *
 20:  * See the enclosed file COPYING for license information (LGPL). If you did not
 21:  * receive this file, see
 22:  * http://www.horde.org/licenses/lgpl21.
 23:  *
 24:  * @category Kolab
 25:  * @package  Kolab_Format
 26:  * @author   Gunnar Wrobel <wrobel@pardus.de>
 27:  * @license  http://www.horde.org/licenses/lgpl21 LGPL 2.1
 28:  * @link     http://www.horde.org/libraries/Horde_Kolab_Format
 29:  */
 30: class Horde_Kolab_Format_Xml_Parser
 31: {
 32:     /**
 33:      * The XML parser.
 34:      *
 35:      * @var DOMDocument
 36:      */
 37:     private $_document;
 38: 
 39:     /**
 40:      * Constructor.
 41:      *
 42:      * @param DOMDocument $document The XML parser.
 43:      */
 44:     public function __construct(DOMDocument $document)
 45:     {
 46:         $this->_document = $document;
 47:         $this->_document->preserveWhiteSpace = false;
 48:         $this->_document->formatOutput       = true;
 49:     }
 50: 
 51:     /**
 52:      * Simply return the DOMDocument without parsing any data.
 53:      *
 54:      * @since Horde_Kolab_Format 1.1.0
 55:      *
 56:      * @return DOMDocument The DOM document.
 57:      */
 58:     public function getDocument()
 59:     {
 60:         return $this->_document;
 61:     }
 62: 
 63:     /**
 64:      * Load an object based on the given XML string.
 65:      *
 66:      * @param string $input   The XML of the message as string.
 67:      * @param array  $options Additional options when parsing the XML.
 68:      * <pre>
 69:      * - relaxed: Relaxed error checking (default: false)
 70:      * </pre>
 71:      *
 72:      * @return DOMDocument The DOM document.
 73:      *
 74:      * @throws Horde_Kolab_Format_Exception If parsing the XML data failed.
 75:      */
 76:     public function parse($input, $options = array())
 77:     {
 78:         if (is_resource($input)) {
 79:             rewind($input);
 80:             $input = stream_get_contents($input);
 81:         }
 82:         try {
 83:             return $this->_parseXml($input, $options);
 84:         } catch (Horde_Kolab_Format_Exception_ParseError $e) {
 85:             if (!function_exists('mb_detect_encoding')) {
 86:                 throw $e;
 87:             }
 88:             /**
 89:              * If the first call does not return successfully this might mean we
 90:              * got an attachment with broken encoding. There are some Kolab
 91:              * client versions in the wild that might have done that. So the
 92:              * next section starts a second attempt by guessing the encoding and
 93:              * trying again.
 94:              */
 95:             if (0 !== strcasecmp(
 96:                     mb_detect_encoding($input, 'UTF-8, ISO-8859-1'), 'UTF-8'
 97:                 )) {
 98:                 $input = mb_convert_encoding($input, 'UTF-8', 'ISO-8859-1');
 99:             }
100:             return $this->_parseXml($input, $options);
101:         }
102:     }
103: 
104:      /**
105:      * Parse the XML string. The root node is returned on success.
106:      *
107:      * @param string $input   The XML of the message as string.
108:      * @param array  $options Additional options when parsing the XML.
109:      *
110:      * @return DOMDocument The DOM document.
111:      *
112:      * @throws Horde_Kolab_Format_Exception If parsing the XML data failed.
113:      */
114:     private function _parseXml($input, $options = array())
115:     {
116:         $result = @$this->_document->loadXML($input);
117:         if (!empty($options['relaxed'])) {
118:             return $this->_document;
119:         }
120:         if (!$result) {
121:             throw new Horde_Kolab_Format_Exception_ParseError($input);
122:         }
123:         if (empty($this->_document->documentElement)) {
124:             throw new Horde_Kolab_Format_Exception_ParseError($input);
125:         }
126:         if (!$this->_document->documentElement->hasChildNodes()) {
127:             throw new Horde_Kolab_Format_Exception_ParseError($input);
128:         }
129:         return $this->_document;
130:     }
131: }
132: