1: <?php
  2: /**
  3:  * Portions Copyright 2005-2007 Zend Technologies USA Inc. (http://www.zend.com)
  4:  * Copyright 2007-2012 Horde LLC (http://www.horde.org/)
  5:  *
  6:  * @author   Chuck Hagenbuch <chuck@horde.org>
  7:  * @license  http://www.horde.org/licenses/bsd BSD
  8:  * @category Horde
  9:  * @package  Feed
 10:  */
 11: 
 12: /**
 13:  * @author   Chuck Hagenbuch <chuck@horde.org>
 14:  * @license  http://www.horde.org/licenses/bsd BSD
 15:  * @category Horde
 16:  * @package  Feed
 17:  */
 18: class Horde_Feed
 19: {
 20:     /**
 21:      * Create a Feed object based on a DOMDocument.
 22:      *
 23:      * @param DOMDocument $doc The DOMDocument object to import.
 24:      *
 25:      * @throws Horde_Feed_Exception
 26:      *
 27:      * @return Horde_Feed_Base The feed object imported from $doc
 28:      */
 29:     public static function create(DOMDocument $doc, $uri = null)
 30:     {
 31:         // Try to find the base feed element or a single <entry> of an
 32:         // Atom feed.
 33:         if ($feed = $doc->getElementsByTagName('feed')->item(0)) {
 34:             // Return an Atom feed.
 35:             return new Horde_Feed_Atom($feed, $uri);
 36:         } elseif ($entry = $doc->getElementsByTagName('entry')->item(0)) {
 37:             // Return an Atom single-entry feed.
 38:             $feeddoc = new DOMDocument($doc->version,
 39:                                        $doc->actualEncoding);
 40:             $feed = $feeddoc->appendChild($feeddoc->createElement('feed'));
 41:             $feed->appendChild($feeddoc->importNode($entry, true));
 42: 
 43:             return new Horde_Feed_Atom($feed, $uri);
 44:         }
 45: 
 46:         // Try to find the base feed element of an RSS feed.
 47:         if ($channel = $doc->getElementsByTagName('channel')->item(0)) {
 48:             // Return an RSS feed.
 49:             return new Horde_Feed_Rss($channel, $uri);
 50:         }
 51: 
 52:         // Try to find an outline element of an OPML blogroll.
 53:         if ($outline = $doc->getElementsByTagName('outline')->item(0)) {
 54:             // Return a blogroll feed.
 55:             return new Horde_Feed_Blogroll($doc->documentElement, $uri);
 56:         }
 57: 
 58:         // $doc does not appear to be a valid feed of the supported
 59:         // types.
 60:         throw new Horde_Feed_Exception('Invalid or unsupported feed format: '
 61:                                        . substr($doc->saveXML(), 0, 80) . '...');
 62:     }
 63: 
 64:     /**
 65:      * Reads a feed represented by $string.
 66:      *
 67:      * @param string $string The XML content of the feed.
 68:      * @param string $uri The feed's URI location, if known.
 69:      *
 70:      * @throws Horde_Feed_Exception
 71:      *
 72:      * @return Horde_Feed_Base
 73:      */
 74:     public static function read($string, $uri = null)
 75:     {
 76:         // Load the feed as a DOMDocument object.
 77:         libxml_use_internal_errors(true);
 78:         $doc = new DOMDocument;
 79:         $doc->recover = true;
 80:         $loaded = $doc->loadXML($string);
 81:         if (!$loaded) {
 82:             $loaded = $doc->loadHTML($string);
 83:             if (!$loaded) {
 84:                 self::_exception('DOMDocument cannot parse XML', libxml_get_last_error());
 85:             }
 86:         }
 87: 
 88:         return self::create($doc);
 89:     }
 90: 
 91:     /**
 92:      * Read a feed located at $uri
 93:      *
 94:      * @param string $uri The URI to fetch the feed from.
 95:      * @param Horde_Http_Client $httpclient The HTTP client to use.
 96:      *
 97:      * @throws Horde_Feed_Exception
 98:      *
 99:      * @return Horde_Feed_Base
100:      */
101:     public static function readUri($uri, Horde_Http_Client $httpclient = null)
102:     {
103:         if (is_null($httpclient)) {
104:             $httpclient = new Horde_Http_Client();
105:         }
106: 
107:         try {
108:             $response = $httpclient->get($uri);
109:         } catch (Horde_Http_Exception $e) {
110:             throw new Horde_Feed_Exception('Error reading feed: ' . $e->getMessage());
111:         }
112:         if ($response->code != 200) {
113:             throw new Horde_Feed_Exception('Unable to read feed, got response code ' . $response->code);
114:         }
115:         $feed = $response->getBody();
116:         return self::read($feed, $uri);
117:     }
118: 
119:     /**
120:      * Read a feed from $filename
121:      *
122:      * @param string $filename The location of the feed file on an accessible
123:      * filesystem or through an available stream wrapper.
124:      *
125:      * @throws Horde_Feed_Exception
126:      *
127:      * @return Horde_Feed_Base
128:      */
129:     public static function readFile($filename)
130:     {
131:         libxml_use_internal_errors(true);
132:         $doc = new DOMDocument;
133:         $doc->recover = true;
134:         $filename = urlencode($filename);
135:         $loaded = $doc->load($filename);
136:         if (!$loaded) {
137:             $loaded = $doc->loadHTMLFile($filename);
138:             if (!$loaded) {
139:                 self::_exception('File could not be read or parsed', libxml_get_last_error());
140:             }
141:         }
142: 
143:         return self::create($doc);
144:     }
145: 
146:     /**
147:      * Builds an exception message from a libXMLError object.
148:      *
149:      * @param string $msg         An error message.
150:      * @param libXMLError $error  An error object.
151:      *
152:      * @throws Horde_Feed_Exception
153:      */
154:     protected static function _exception($msg, $error)
155:     {
156:         if ($error) {
157:             $msg .= ': ' . $error->message;
158:             if ($error->file) {
159:                 $msg .= sprintf(' in file %s, line %d, column %d',
160:                                 $error->file, $error->line, $error->column);
161:             }
162:         }
163:         throw new Horde_Feed_Exception($msg);
164:     }
165: }
166: