1: <?php
  2: 
  3: /*
  4:  * This file is part of SwiftMailer.
  5:  * (c) 2004-2009 Chris Corbyn
  6:  *
  7:  * For the full copyright and license information, please view the LICENSE
  8:  * file that was distributed with this source code.
  9:  */
 10: 
 11: /**
 12:  * Defines the grammar to use for validation, implements the RFC 2822 (and friends) ABNF grammar definitions.
 13:  * @package Swift
 14:  * @subpackage Mime
 15:  * @author Fabien Potencier
 16:  * @author Chris Corbyn
 17:  */
 18: class Swift_Mime_Grammar
 19: {
 20:     /**
 21:      * Special characters used in the syntax which need to be escaped.
 22:      * @var string[]
 23:      * @access private
 24:      */
 25:     private static $_specials = array();
 26: 
 27:     /**
 28:      * Tokens defined in RFC 2822 (and some related RFCs).
 29:      * @var string[]
 30:      * @access private
 31:      */
 32:     private static $_grammar = array();
 33: 
 34:     /**
 35:      * Initialize some RFC 2822 (and friends) ABNF grammar definitions.
 36:      * @access protected
 37:      */
 38:     public function __construct()
 39:     {
 40:         $this->init();
 41:     }
 42: 
 43:     public function __wakeup()
 44:     {
 45:         $this->init();
 46:     }
 47: 
 48:     protected function init()
 49:     {
 50:         if (count(self::$_specials) > 0) {
 51:             return;
 52:         }
 53: 
 54:         self::$_specials = array(
 55:             '(', ')', '<', '>', '[', ']',
 56:             ':', ';', '@', ',', '.', '"'
 57:             );
 58: 
 59:         /*** Refer to RFC 2822 for ABNF grammar ***/
 60: 
 61:         //All basic building blocks
 62:         self::$_grammar['NO-WS-CTL'] = '[\x01-\x08\x0B\x0C\x0E-\x19\x7F]';
 63:         self::$_grammar['WSP'] = '[ \t]';
 64:         self::$_grammar['CRLF'] = '(?:\r\n)';
 65:         self::$_grammar['FWS'] = '(?:(?:' . self::$_grammar['WSP'] . '*' .
 66:                 self::$_grammar['CRLF'] . ')?' . self::$_grammar['WSP'] . ')';
 67:         self::$_grammar['text'] = '[\x00-\x08\x0B\x0C\x0E-\x7F]';
 68:         self::$_grammar['quoted-pair'] = '(?:\\\\' . self::$_grammar['text'] . ')';
 69:         self::$_grammar['ctext'] = '(?:' . self::$_grammar['NO-WS-CTL'] .
 70:                 '|[\x21-\x27\x2A-\x5B\x5D-\x7E])';
 71:         //Uses recursive PCRE (?1) -- could be a weak point??
 72:         self::$_grammar['ccontent'] = '(?:' . self::$_grammar['ctext'] . '|' .
 73:                 self::$_grammar['quoted-pair'] . '|(?1))';
 74:         self::$_grammar['comment'] = '(\((?:' . self::$_grammar['FWS'] . '|' .
 75:                 self::$_grammar['ccontent']. ')*' . self::$_grammar['FWS'] . '?\))';
 76:         self::$_grammar['CFWS'] = '(?:(?:' . self::$_grammar['FWS'] . '?' .
 77:                 self::$_grammar['comment'] . ')*(?:(?:' . self::$_grammar['FWS'] . '?' .
 78:                 self::$_grammar['comment'] . ')|' . self::$_grammar['FWS'] . '))';
 79:         self::$_grammar['qtext'] = '(?:' . self::$_grammar['NO-WS-CTL'] .
 80:                 '|[\x21\x23-\x5B\x5D-\x7E])';
 81:         self::$_grammar['qcontent'] = '(?:' . self::$_grammar['qtext'] . '|' .
 82:                 self::$_grammar['quoted-pair'] . ')';
 83:         self::$_grammar['quoted-string'] = '(?:' . self::$_grammar['CFWS'] . '?"' .
 84:                 '(' . self::$_grammar['FWS'] . '?' . self::$_grammar['qcontent'] . ')*' .
 85:                 self::$_grammar['FWS'] . '?"' . self::$_grammar['CFWS'] . '?)';
 86:         self::$_grammar['atext'] = '[a-zA-Z0-9!#\$%&\'\*\+\-\/=\?\^_`\{\}\|~]';
 87:         self::$_grammar['atom'] = '(?:' . self::$_grammar['CFWS'] . '?' .
 88:                 self::$_grammar['atext'] . '+' . self::$_grammar['CFWS'] . '?)';
 89:         self::$_grammar['dot-atom-text'] = '(?:' . self::$_grammar['atext'] . '+' .
 90:                 '(\.' . self::$_grammar['atext'] . '+)*)';
 91:         self::$_grammar['dot-atom'] = '(?:' . self::$_grammar['CFWS'] . '?' .
 92:                 self::$_grammar['dot-atom-text'] . '+' . self::$_grammar['CFWS'] . '?)';
 93:         self::$_grammar['word'] = '(?:' . self::$_grammar['atom'] . '|' .
 94:                 self::$_grammar['quoted-string'] . ')';
 95:         self::$_grammar['phrase'] = '(?:' . self::$_grammar['word'] . '+?)';
 96:         self::$_grammar['no-fold-quote'] = '(?:"(?:' . self::$_grammar['qtext'] .
 97:                 '|' . self::$_grammar['quoted-pair'] . ')*")';
 98:         self::$_grammar['dtext'] = '(?:' . self::$_grammar['NO-WS-CTL'] .
 99:                 '|[\x21-\x5A\x5E-\x7E])';
100:         self::$_grammar['no-fold-literal'] = '(?:\[(?:' . self::$_grammar['dtext'] .
101:                 '|' . self::$_grammar['quoted-pair'] . ')*\])';
102: 
103:         //Message IDs
104:         self::$_grammar['id-left'] = '(?:' . self::$_grammar['dot-atom-text'] . '|' .
105:                 self::$_grammar['no-fold-quote'] . ')';
106:         self::$_grammar['id-right'] = '(?:' . self::$_grammar['dot-atom-text'] . '|' .
107:                 self::$_grammar['no-fold-literal'] . ')';
108: 
109:         //Addresses, mailboxes and paths
110:         self::$_grammar['local-part'] = '(?:' . self::$_grammar['dot-atom'] . '|' .
111:                 self::$_grammar['quoted-string'] . ')';
112:         self::$_grammar['dcontent'] = '(?:' . self::$_grammar['dtext'] . '|' .
113:                 self::$_grammar['quoted-pair'] . ')';
114:         self::$_grammar['domain-literal'] = '(?:' . self::$_grammar['CFWS'] . '?\[(' .
115:                 self::$_grammar['FWS'] . '?' . self::$_grammar['dcontent'] . ')*?' .
116:                 self::$_grammar['FWS'] . '?\]' . self::$_grammar['CFWS'] . '?)';
117:         self::$_grammar['domain'] = '(?:' . self::$_grammar['dot-atom'] . '|' .
118:                 self::$_grammar['domain-literal'] . ')';
119:         self::$_grammar['addr-spec'] = '(?:' . self::$_grammar['local-part'] . '@' .
120:                 self::$_grammar['domain'] . ')';
121:     }
122: 
123:     /**
124:      * Get the grammar defined for $name token.
125:      * @param  string $name execatly as written in the RFC
126:      * @return string
127:      */
128:     public function getDefinition($name)
129:     {
130:         if (array_key_exists($name, self::$_grammar)) {
131:             return self::$_grammar[$name];
132:         } else {
133:             throw new Swift_RfcComplianceException(
134:                 "No such grammar '" . $name . "' defined."
135:                 );
136:         }
137:     }
138: 
139:     /**
140:      * Returns the tokens defined in RFC 2822 (and some related RFCs).
141:      * @return array
142:      */
143:     public function getGrammarDefinitions()
144:     {
145:         return self::$_grammar;
146:     }
147: 
148:     /**
149:      * Returns the current special characters used in the syntax which need to be escaped.
150:      * @return array
151:      */
152:     public function getSpecials()
153:     {
154:         return self::$_specials;
155:     }
156: 
157:     /**
158:      * Escape special characters in a string (convert to quoted-pairs).
159:      * @param  string   $token
160:      * @param  string[] $include additonal chars to escape
161:      * @param  string[] $exclude chars from escaping
162:      * @return string
163:      */
164:     public function escapeSpecials($token, $include = array(), $exclude = array())
165:     {
166:         foreach (
167:             array_merge(array('\\'), array_diff(self::$_specials, $exclude), $include) as $char)
168:         {
169:             $token = str_replace($char, '\\' . $char, $token);
170:         }
171: 
172:         return $token;
173:     }
174: }
175: