| 1 |
<?php |
| 2 |
/** |
| 3 |
* class FeedFinder: find likely feeds using autodetection and/or guesswork |
| 4 |
* @version 2010.0622 |
| 5 |
* @uses SimplePie_Misc |
| 6 |
*/ |
| 7 |
|
| 8 |
if (!class_exists('SimplePie')) : |
| 9 |
require_once(ABSPATH . WPINC . '/class-simplepie.php'); |
| 10 |
endif; |
| 11 |
require_once(dirname(__FILE__).'/feedwordpresshtml.class.php'); |
| 12 |
|
| 13 |
class FeedFinder { |
| 14 |
var $uri = NULL; |
| 15 |
var $_cache_uri = NULL; |
| 16 |
|
| 17 |
var $verify = FALSE; |
| 18 |
var $fallbacks = 3; |
| 19 |
|
| 20 |
var $_response = NULL; |
| 21 |
var $_data = NULL; |
| 22 |
var $_error = NULL; |
| 23 |
var $_head = NULL; |
| 24 |
|
| 25 |
# -- Recognition patterns |
| 26 |
var $_feed_types = array( |
| 27 |
'application/rss+xml', |
| 28 |
'text/xml', |
| 29 |
'application/atom+xml', |
| 30 |
'application/x.atom+xml', |
| 31 |
'application/x-atom+xml' |
| 32 |
); |
| 33 |
var $_feed_markers = array('\\<feed', '\\<rss', 'xmlns="http://purl.org/rss/1.0'); |
| 34 |
var $_html_markers = array('\\<html'); |
| 35 |
var $_opml_markers = array('\\<opml', '\\<outline'); |
| 36 |
var $_obvious_feed_url = array('[./]rss', '[./]rdf', '[./]atom', '[./]feed', '\.xml'); |
| 37 |
var $_maybe_feed_url = array ('rss', 'rdf', 'atom', 'feed', 'xml'); |
| 38 |
|
| 39 |
function FeedFinder ($uri = NULL, $verify = TRUE, $fallbacks = 3) { |
| 40 |
$this->uri = $uri; $this->verify = $verify; |
| 41 |
$this->fallbacks = $fallbacks; |
| 42 |
} /* FeedFinder::FeedFinder () */ |
| 43 |
|
| 44 |
function find ($uri = NULL) { |
| 45 |
$ret = array (); |
| 46 |
if (!is_null($this->data($uri))) : |
| 47 |
if ($this->is_opml($uri)) : |
| 48 |
$href = $this->_opml_rss_uris(); |
| 49 |
else : |
| 50 |
if ($this->is_feed($uri)) : |
| 51 |
$href = array($this->uri); |
| 52 |
else : |
| 53 |
// Assume that we have HTML or XHTML (even if we don't, who's it gonna hurt?) |
| 54 |
// Autodiscovery is the preferred method |
| 55 |
$href = $this->_link_rel_feeds(); |
| 56 |
|
| 57 |
// ... but we'll also take the little orange buttons |
| 58 |
if ($this->fallbacks > 0) : |
| 59 |
$href = array_merge($href, $this->_a_href_feeds(TRUE)); |
| 60 |
endif; |
| 61 |
|
| 62 |
// If all that failed, look harder |
| 63 |
if ($this->fallbacks > 1) : |
| 64 |
if (count($href) == 0) : |
| 65 |
$href = $this->_a_href_feeds(FALSE); |
| 66 |
endif; |
| 67 |
endif; |
| 68 |
|
| 69 |
// Our search may turn up duplicate URIs. We only need to do any given URI once. |
| 70 |
// Props to Camilo <http://projects.radgeek.com/2008/12/14/feedwordpress-20081214/#comment-20090122160414> |
| 71 |
$href = array_unique($href); |
| 72 |
endif; |
| 73 |
|
| 74 |
// Try some clever URL little tricks before we go |
| 75 |
if ($this->fallbacks > 2) : |
| 76 |
$href = array_merge($href, $this->_url_manipulation_feeds()); |
| 77 |
endif; |
| 78 |
endif; |
| 79 |
|
| 80 |
$href = array_unique($href); |
| 81 |
|
| 82 |
// Verify feeds and resolve relative URIs |
| 83 |
foreach ($href as $u) : |
| 84 |
$the_uri = SimplePie_Misc::absolutize_url($u, $this->uri); |
| 85 |
if ($this->verify and ($u != $this->uri and $the_uri != $this->uri)) : |
| 86 |
$feed = new FeedFinder($the_uri); |
| 87 |
if ($feed->is_feed()) : $ret[] = $the_uri; endif; |
| 88 |
unset($feed); |
| 89 |
else : |
| 90 |
$ret[] = $the_uri; |
| 91 |
endif; |
| 92 |
endforeach; |
| 93 |
endif; |
| 94 |
|
| 95 |
return array_values($ret); |
| 96 |
} /* FeedFinder::find () */ |
| 97 |
|
| 98 |
function data ($uri = NULL) { |
| 99 |
$this->_get($uri); |
| 100 |
return $this->_data; |
| 101 |
} /* FeedFinder::data () */ |
| 102 |
|
| 103 |
function upload_data ($data) { |
| 104 |
$this->uri = 'tag:localhost'; |
| 105 |
$this->_data = $data; |
| 106 |
} /* FeedFinder::upload_data () */ |
| 107 |
|
| 108 |
function status ($uri = NULL) { |
| 109 |
$this->_get($uri); |
| 110 |
|
| 111 |
if (!is_wp_error($this->_response) and isset($this->_response['response']['code'])) : |
| 112 |
$ret = $this->_response['response']['code']; |
| 113 |
else : |
| 114 |
$ret = NULL; |
| 115 |
endif; |
| 116 |
return $ret; |
| 117 |
} |
| 118 |
|
| 119 |
function error ($index = NULL) { |
| 120 |
$message = NULL; |
| 121 |
if (count($this->_error) > 0) : |
| 122 |
if (is_scalar($index) and !is_null($index)) : |
| 123 |
$message = $this->_error[$index]; |
| 124 |
else : |
| 125 |
$message = implode(" / ", $this->_error)."\n"; |
| 126 |
endif; |
| 127 |
endif; |
| 128 |
return $message; |
| 129 |
} |
| 130 |
|
| 131 |
function is_feed ($uri = NULL) { |
| 132 |
$data = $this->data($uri); |
| 133 |
|
| 134 |
return ( |
| 135 |
preg_match ( |
| 136 |
"\007(".implode('|',$this->_feed_markers).")\007i", |
| 137 |
$data |
| 138 |
) and !preg_match ( |
| 139 |
"\007(".implode('|',$this->_html_markers).")\007i", |
| 140 |
$data |
| 141 |
) |
| 142 |
); |
| 143 |
} /* FeedFinder::is_feed () */ |
| 144 |
|
| 145 |
function is_opml ($uri = NULL) { |
| 146 |
$data = $this->data($uri); |
| 147 |
return ( |
| 148 |
preg_match ( |
| 149 |
"\007(".implode('|',$this->_opml_markers).")\007i", |
| 150 |
$data |
| 151 |
) |
| 152 |
); |
| 153 |
} /* FeedFinder::is_opml () */ |
| 154 |
|
| 155 |
# --- Private methods --- |
| 156 |
function _get ($uri = NULL) { |
| 157 |
if ($uri) $this->uri = $uri; |
| 158 |
|
| 159 |
// Is the result not yet cached? |
| 160 |
if ($this->uri != 'tag:localhost' and $this->_cache_uri !== $this->uri) : |
| 161 |
$headers['Connection'] = 'close'; |
| 162 |
$headers['Accept'] = 'application/atom+xml application/rdf+xml application/rss+xml application/xml text/html */*'; |
| 163 |
$headers['User-Agent'] = 'feedfinder/1.2 (compatible; PHP FeedFinder) +http://projects.radgeek.com/feedwordpress'; |
| 164 |
|
| 165 |
// Use WordPress API function |
| 166 |
$client = wp_remote_request($this->uri, array( |
| 167 |
'headers' => $headers, |
| 168 |
'timeout' => FEEDWORDPRESS_FETCH_TIME_OUT, |
| 169 |
)); |
| 170 |
|
| 171 |
$this->_response = $client; |
| 172 |
if (is_wp_error($client)) : |
| 173 |
$this->_data = NULL; |
| 174 |
$this->_error = $client->get_error_messages(); |
| 175 |
else : |
| 176 |
$this->_data = $client['body']; |
| 177 |
$this->_error = NULL; |
| 178 |
endif; |
| 179 |
|
| 180 |
// Kilroy was here |
| 181 |
$this->_cache_uri = $this->uri; |
| 182 |
endif; |
| 183 |
} /* FeedFinder::_get () */ |
| 184 |
|
| 185 |
function _opml_rss_uris () { |
| 186 |
// Really we should parse the XML and use the structure to |
| 187 |
// return something intelligent to programs that want to use it |
| 188 |
// Oh babe! maybe some day... |
| 189 |
|
| 190 |
$opml = $this->data(); |
| 191 |
|
| 192 |
$rx = FeedWordPressHTML::attributeRegex('outline', 'xmlUrl'); |
| 193 |
if (preg_match_all($rx, $opml, $matches, PREG_SET_ORDER)) : |
| 194 |
foreach ($matches as $m) : |
| 195 |
$match = FeedWordPressHTML::attributeMatch($m); |
| 196 |
$r[] = $match['value']; |
| 197 |
endforeach; |
| 198 |
endif; |
| 199 |
return $r; |
| 200 |
} /* FeedFinder::_opml_rss_uris () */ |
| 201 |
|
| 202 |
function _link_rel_feeds () { |
| 203 |
$links = $this->_tags('link'); |
| 204 |
$link_count = count($links); |
| 205 |
|
| 206 |
// now figure out which one points to the RSS file |
| 207 |
$href = array (); |
| 208 |
for ($n=0; $n<$link_count; $n++) { |
| 209 |
if (strtolower($links[$n]['rel']) == 'alternate') { |
| 210 |
if (in_array(strtolower($links[$n]['type']), $this->_feed_types)) { |
| 211 |
$href[] = $links[$n]['href']; |
| 212 |
} /* if */ |
| 213 |
} /* if */ |
| 214 |
} /* for */ |
| 215 |
return $href; |
| 216 |
} /* FeedFinder::_link_rel_feeds () */ |
| 217 |
|
| 218 |
function _a_href_feeds ($obvious = TRUE) { |
| 219 |
$pattern = ($obvious ? $this->_obvious_feed_url : $this->_maybe_feed_url); |
| 220 |
|
| 221 |
$links = $this->_tags('a'); |
| 222 |
$link_count = count($links); |
| 223 |
|
| 224 |
// now figure out which one points to the RSS file |
| 225 |
$href = array (); |
| 226 |
for ($n=0; $n<$link_count; $n++) { |
| 227 |
if (preg_match("\007(".implode('|',$pattern).")\007i", $links[$n]['href'])) { |
| 228 |
$href[] = $links[$n]['href']; |
| 229 |
} /* if */ |
| 230 |
} /* for */ |
| 231 |
return $href; |
| 232 |
} /* FeedFinder::_link_a_href_feeds () */ |
| 233 |
|
| 234 |
function _url_manipulation_feeds () { |
| 235 |
$href = array(); |
| 236 |
|
| 237 |
// check for HTTP GET parameters that look feed-like. |
| 238 |
$bits = parse_url($this->uri); |
| 239 |
foreach (array('rss', 'rss2', 'atom', 'rdf') as $format) : |
| 240 |
if (isset($bits['query']) and (strlen($bits['query']) > 0)) : |
| 241 |
$newQuery = preg_replace('/([?&=])(rss2?|atom|rdf)/i', '$1'.$format, $bits['query']); |
| 242 |
else : |
| 243 |
$newQuery = NULL; |
| 244 |
endif; |
| 245 |
|
| 246 |
if (isset($bits['path']) and (strlen($bits['path']) > 0)) : |
| 247 |
$newPath = preg_replace('!([/.])(rss2?|atom|rdf)!i', '$1'.$format, $bits['path']); |
| 248 |
else : |
| 249 |
$newPath = NULL; |
| 250 |
endif; |
| 251 |
|
| 252 |
// Reassemble, check and return |
| 253 |
$credentials = ''; |
| 254 |
if (isset($bits['user'])) : |
| 255 |
$credentials = $bits['user']; |
| 256 |
if (isset($bits['pass'])) : |
| 257 |
$credentials .= ':'.$bits['pass']; |
| 258 |
endif; |
| 259 |
$credentials .= '@'; |
| 260 |
endif; |
| 261 |
|
| 262 |
// Variations on a theme |
| 263 |
$newUrl[0] = ('' |
| 264 |
.(isset($bits['scheme']) ? $bits['scheme'].':' : '') |
| 265 |
.(isset($bits['host']) ? '//'.$credentials.$bits['host'] : '') |
| 266 |
.(!is_null($newPath) ? $newPath : '') |
| 267 |
.(!is_null($newQuery) ? '?'.$newQuery : '') |
| 268 |
.(isset($bits['fragment']) ? '#'.$bits['fragment'] : '') |
| 269 |
); |
| 270 |
$newUrl[1] = ('' |
| 271 |
.(isset($bits['scheme']) ? $bits['scheme'].':' : '') |
| 272 |
.(isset($bits['host']) ? '//'.$credentials.$bits['host'] : '') |
| 273 |
.(!is_null($newPath) ? $newPath : '') |
| 274 |
.(isset($bits['query']) ? '?'.$bits['query'] : '') |
| 275 |
.(isset($bits['fragment']) ? '#'.$bits['fragment'] : '') |
| 276 |
); |
| 277 |
$newUrl[2] = ('' |
| 278 |
.(isset($bits['scheme']) ? $bits['scheme'].':' : '') |
| 279 |
.(isset($bits['host']) ? '//'.$credentials.$bits['host'] : '') |
| 280 |
.(isset($bits['path']) ? $bits['path'] : '') |
| 281 |
.(!is_null($newQuery) ? '?'.$newQuery : '') |
| 282 |
.(isset($bits['fragment']) ? '#'.$bits['fragment'] : '') |
| 283 |
); |
| 284 |
$href = array_merge($href, $newUrl); |
| 285 |
endforeach; |
| 286 |
return array_unique($href); |
| 287 |
} /* FeedFinder::_url_manipulation_feeds () */ |
| 288 |
|
| 289 |
function _tags ($tag) { |
| 290 |
$html = $this->data(); |
| 291 |
|
| 292 |
// search through the HTML, save all <link> tags |
| 293 |
// and store each link's attributes in an associative array |
| 294 |
preg_match_all('/<'.$tag.'\s+(.*?)\s*\/?>/si', $html, $matches); |
| 295 |
$links = $matches[1]; |
| 296 |
$ret = array(); |
| 297 |
$link_count = count($links); |
| 298 |
for ($n=0; $n<$link_count; $n++) { |
| 299 |
$attributes = preg_split('/\s+/s', $links[$n]); |
| 300 |
foreach($attributes as $attribute) { |
| 301 |
$att = preg_split('/\s*=\s*/s', $attribute, 2); |
| 302 |
if (isset($att[1])) { |
| 303 |
$att[1] = preg_replace('/([\'"]?)(.*)\1/', '$2', $att[1]); |
| 304 |
$final_link[strtolower($att[0])] = $att[1]; |
| 305 |
} /* if */ |
| 306 |
} /* foreach */ |
| 307 |
$ret[$n] = $final_link; |
| 308 |
} /* for */ |
| 309 |
return $ret; |
| 310 |
} /* FeedFinder::_tags () */ |
| 311 |
} /* class FeedFinder */ |
| 312 |
|
| 313 |
|
