← All changes | feedfinder.class.php +211 -43 2010.0528 → 2013.0504 View file →

		@@ -1,15 +1,24 @@
1	1	<?php
2		-################################################################################
3		-## class FeedFinder: find likely feeds using autodetection and/or guesswork ####
4		-################################################################################
	2	+/**
	3	+ * class FeedFinder: find likely feeds using autodetection and/or guesswork
	4	+ * @version 2010.0622
	5	+ * @uses SimplePie_Misc
	6	+ */
5	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	+
6	13	class FeedFinder {
7	14	var $uri = NULL;
	15	+ var $credentials = NULL;
8	16	var $_cache_uri = NULL;
9	17
10	18	var $verify = FALSE;
11		-
	19	+ var $fallbacks = 3;
	20	+
12	21	var $_response = NULL;
13	22	var $_data = NULL;
14	23	var $_error = NULL;
15	24	var $_head = NULL;
		@@ -23,56 +32,126 @@
23	32	'application/x-atom+xml'
24	33	);
25	34	var $_feed_markers = array('\\<feed', '\\<rss', 'xmlns="http://purl.org/rss/1.0');
26	35	var $_html_markers = array('\\<html');
	36	+ var $_opml_markers = array('\\<opml', '\\<outline');
27	37	var $_obvious_feed_url = array('[./]rss', '[./]rdf', '[./]atom', '[./]feed', '\.xml');
28	38	var $_maybe_feed_url = array ('rss', 'rdf', 'atom', 'feed', 'xml');
29	39
30		- function FeedFinder ($uri = NULL, $verify = TRUE) {
	40	+ function FeedFinder ($uri = NULL, $params = array(), $fallbacks = 3) {
	41	+ if (is_bool($params)) :
	42	+ $params = array("verify" => $params);
	43	+ endif;
	44	+
	45	+ $params = wp_parse_args($params, array(
	46	+ "verify" => true,
	47	+ "authentication" => NULL,
	48	+ "username" => NULL,
	49	+ "password" => NULL,
	50	+ ));
	51	+ $verify = $params['verify'];
	52	+ $this->credentials = array(
	53	+ "authentication" => $params['authentication'],
	54	+ "username" => $params['username'],
	55	+ "password" => $params['password'],
	56	+ );
	57	+
31	58	$this->uri = $uri; $this->verify = $verify;
	59	+ $this->fallbacks = $fallbacks;
32	60	} /* FeedFinder::FeedFinder () */
33	61
34		- function find ($uri = NULL) {
	62	+ function find ($uri = NULL, $params = array()) {
	63	+ $params = wp_parse_args($params, array( // Defaults
	64	+ "authentication" => -1,
	65	+ "username" => NULL,
	66	+ "password" => NULL,
	67	+ ));
	68	+
	69	+ // Equivalents
	70	+ if ($params['authentication']=='-') :
	71	+ $params['authentication'] = NULL;
	72	+ $params['username'] = NULL;
	73	+ $params['password'] = NULL;
	74	+ endif;
	75	+
	76	+ // Set/reset
	77	+ if ($params['authentication'] != -1) :
	78	+ $this->credentials = array(
	79	+ "authentication" => $params['authentication'],
	80	+ "username" => $params['username'],
	81	+ "password" => $params['password'],
	82	+ );
	83	+ endif;
	84	+
35	85	$ret = array ();
36		- if (!is_null($this->data($uri))) {
37		- if ($this->is_feed($uri)) {
38		- $ret = array($this->uri);
39		- } else {
40		- // Assume that we have HTML or XHTML (even if we don't, who's it gonna hurt?)
41		- // Autodiscovery is the preferred method
42		- $href = $this->_link_rel_feeds();
43		-
44		- // ... but we'll also take the little orange buttons
45		- $href = array_merge($href, $this->_a_href_feeds(TRUE));
46		-
47		- // If all that failed, look harder
48		- if (count($href) == 0) $href = $this->_a_href_feeds(FALSE);
	86	+ if (!is_null($this->data($uri))) :
	87	+ if ($this->is_opml($uri)) :
	88	+ $href = $this->_opml_rss_uris();
	89	+ else :
	90	+ if ($this->is_feed($uri)) :
	91	+ $href = array($this->uri);
	92	+ else :
	93	+ // Assume that we have HTML or XHTML (even if we don't, who's
	94	+ // it gonna hurt?) Autodiscovery is the preferred method.
	95	+ $href = $this->_link_rel_feeds();
	96	+
	97	+ // ... but we'll also take the little orange buttons
	98	+ if ($this->fallbacks > 0) :
	99	+ $href = array_merge($href, $this->_a_href_feeds(TRUE));
	100	+ endif;
	101	+
	102	+ // If all that failed, look harder
	103	+ if ($this->fallbacks > 1) :
	104	+ if (count($href) == 0) :
	105	+ $href = $this->_a_href_feeds(FALSE);
	106	+ endif;
	107	+ endif;
	108	+
	109	+ // Our search may turn up duplicate URIs. We only need to do
	110	+ // any given URI once. Props to Camilo <http://projects.radgeek.com/2008/12/14/feedwordpress-20081214/#comment-20090122160414>
	111	+ $href = array_unique($href);
	112	+ endif;
49	113
50		- // Our search may turn up duplicate URIs. We only need to do any given URI once.
51		- // Props to Camilo <http://projects.radgeek.com/2008/12/14/feedwordpress-20081214/#comment-20090122160414>
52		- $href = array_unique($href);
	114	+ // Try some clever URL little tricks before we go
	115	+ if ($this->fallbacks > 2) :
	116	+ $href = array_merge($href, $this->_url_manipulation_feeds());
	117	+ endif;
	118	+ endif;
	119	+
	120	+ $href = array_unique($href);
53	121
54		- // Verify feeds and resolve relative URIs
55		- foreach ($href as $u) {
56		- $the_uri = Relative_URI::resolve($u, $this->uri);
57		- if ($this->verify) {
58		- $feed = new FeedFinder($the_uri);
59		- if ($feed->is_feed()) $ret[] = $the_uri;
60		- unset($feed);
61		- } else {
62		- $ret[] = $the_uri;
63		- }
64		- } /* foreach */
65		- } /* if */
66		- } /* if */
67		- return array_unique($ret);
	122	+ // Verify feeds and resolve relative URIs
	123	+ foreach ($href as $u) :
	124	+ $the_uri = SimplePie_Misc::absolutize_url($u, $this->uri);
	125	+ if ($this->verify and ($u != $this->uri and $the_uri != $this->uri)) :
	126	+ $feed = new FeedFinder($the_uri, $this->credentials);
	127	+ if ($feed->is_feed()) : $ret[] = $the_uri; endif;
	128	+ unset($feed);
	129	+ else :
	130	+ $ret[] = $the_uri;
	131	+ endif;
	132	+ endforeach;
	133	+ endif;
	134	+
	135	+ if ($this->is_401($uri)) :
	136	+ $ret = array_merge(array(
	137	+ new WP_Error('http_request_failed', '401 Not authorized', array("uri" => $this->uri, "status" => 401)),
	138	+ ), $ret);
	139	+ endif;
	140	+
	141	+ return array_values($ret);
68	142	} /* FeedFinder::find () */
69	143
70	144	function data ($uri = NULL) {
71	145	$this->_get($uri);
72	146	return $this->_data;
73		- }
	147	+ } /* FeedFinder::data () */
74	148
	149	+ function upload_data ($data) {
	150	+ $this->uri = 'tag:localhost';
	151	+ $this->_data = $data;
	152	+ } /* FeedFinder::upload_data () */
	153	+
75	154	function status ($uri = NULL) {
76	155	$this->_get($uri);
77	156
78	157	if (!is_wp_error($this->_response) and isset($this->_response['response']['code'])) :
		@@ -94,8 +173,12 @@
94	173	endif;
95	174	return $message;
96	175	}
97	176
	177	+ function is_401 ($uri = NULL) {
	178	+ return (intval($this->status($uri))==401);
	179	+ } /* FeedFinder::is_401 () */
	180	+
98	181	function is_feed ($uri = NULL) {
99	182	$data = $this->data($uri);
100	183
101	184	return (
		@@ -108,22 +191,35 @@
108	191	)
109	192	);
110	193	} /* FeedFinder::is_feed () */
111	194
	195	+ function is_opml ($uri = NULL) {
	196	+ $data = $this->data($uri);
	197	+ return (
	198	+ preg_match (
	199	+ "\007(".implode('|',$this->_opml_markers).")\007i",
	200	+ $data
	201	+ )
	202	+ );
	203	+ } /* FeedFinder::is_opml () */
	204	+
112	205	# --- Private methods ---
113	206	function _get ($uri = NULL) {
114	207	if ($uri) $this->uri = $uri;
115	208
116	209	// Is the result not yet cached?
117		- if ($this->_cache_uri !== $this->uri) :
	210	+ if ($this->uri != 'tag:localhost' and $this->_cache_uri !== $this->uri) :
118	211	$headers['Connection'] = 'close';
119	212	$headers['Accept'] = 'application/atom+xml application/rdf+xml application/rss+xml application/xml text/html */*';
120	213	$headers['User-Agent'] = 'feedfinder/1.2 (compatible; PHP FeedFinder) +http://projects.radgeek.com/feedwordpress';
121	214
122	215	// Use WordPress API function
123		- $client = wp_remote_request($this->uri, array(
	216	+ $client = wp_remote_request($this->uri, array_merge(
	217	+ $this->credentials,
	218	+ array(
124	219	'headers' => $headers,
125		- 'timeout' => FEEDWORDPRESS_FETCH_TIME_OUT,
	220	+ 'timeout' => FeedWordPress::fetch_timeout(),
	221	+ )
126	222	));
127	223
128	224	$this->_response = $client;
129	225	if (is_wp_error($client)) :
		@@ -138,8 +234,25 @@
138	234	$this->_cache_uri = $this->uri;
139	235	endif;
140	236	} /* FeedFinder::_get () */
141	237
	238	+ function _opml_rss_uris () {
	239	+ // Really we should parse the XML and use the structure to
	240	+ // return something intelligent to programs that want to use it
	241	+ // Oh babe! maybe some day...
	242	+
	243	+ $opml = $this->data();
	244	+
	245	+ $rx = FeedWordPressHTML::attributeRegex('outline', 'xmlUrl');
	246	+ if (preg_match_all($rx, $opml, $matches, PREG_SET_ORDER)) :
	247	+ foreach ($matches as $m) :
	248	+ $match = FeedWordPressHTML::attributeMatch($m);
	249	+ $r[] = $match['value'];
	250	+ endforeach;
	251	+ endif;
	252	+ return $r;
	253	+ } /* FeedFinder::_opml_rss_uris () */
	254	+
142	255	function _link_rel_feeds () {
143	256	$links = $this->_tags('link');
144	257	$link_count = count($links);
145	258
		@@ -152,9 +265,9 @@
152	265	} /* if */
153	266	} /* if */
154	267	} /* for */
155	268	return $href;
156		- }
	269	+ } /* FeedFinder::_link_rel_feeds () */
157	270
158	271	function _a_href_feeds ($obvious = TRUE) {
159	272	$pattern = ($obvious ? $this->_obvious_feed_url : $this->_maybe_feed_url);
160	273
		@@ -168,10 +281,65 @@
168	281	$href[] = $links[$n]['href'];
169	282	} /* if */
170	283	} /* for */
171	284	return $href;
172		- }
	285	+ } /* FeedFinder::_link_a_href_feeds () */
173	286
	287	+ function _url_manipulation_feeds () {
	288	+ $href = array();
	289	+
	290	+ // check for HTTP GET parameters that look feed-like.
	291	+ $bits = parse_url($this->uri);
	292	+ foreach (array('rss', 'rss2', 'atom', 'rdf') as $format) :
	293	+ if (isset($bits['query']) and (strlen($bits['query']) > 0)) :
	294	+ $newQuery = preg_replace('/([?&=])(rss2?|atom|rdf)/i', '$1'.$format, $bits['query']);
	295	+ else :
	296	+ $newQuery = NULL;
	297	+ endif;
	298	+
	299	+ if (isset($bits['path']) and (strlen($bits['path']) > 0)) :
	300	+ $newPath = preg_replace('!([/.])(rss2?|atom|rdf)!i', '$1'.$format, $bits['path']);
	301	+ else :
	302	+ $newPath = NULL;
	303	+ endif;
	304	+
	305	+ // Reassemble, check and return
	306	+ $credentials = '';
	307	+ if (isset($bits['user'])) :
	308	+ $credentials = $bits['user'];
	309	+ if (isset($bits['pass'])) :
	310	+ $credentials .= ':'.$bits['pass'];
	311	+ endif;
	312	+ $credentials .= '@';
	313	+ endif;
	314	+
	315	+ // Variations on a theme
	316	+ $newUrl[0] = (''
	317	+ .(isset($bits['scheme']) ? $bits['scheme'].':' : '')
	318	+ .(isset($bits['host']) ? '//'.$credentials.$bits['host'] : '')
	319	+ .(!is_null($newPath) ? $newPath : '')
	320	+ .(!is_null($newQuery) ? '?'.$newQuery : '')
	321	+ .(isset($bits['fragment']) ? '#'.$bits['fragment'] : '')
	322	+ );
	323	+ $newUrl[1] = (''
	324	+ .(isset($bits['scheme']) ? $bits['scheme'].':' : '')
	325	+ .(isset($bits['host']) ? '//'.$credentials.$bits['host'] : '')
	326	+ .(!is_null($newPath) ? $newPath : '')
	327	+ .(isset($bits['query']) ? '?'.$bits['query'] : '')
	328	+ .(isset($bits['fragment']) ? '#'.$bits['fragment'] : '')
	329	+ );
	330	+ $newUrl[2] = (''
	331	+ .(isset($bits['scheme']) ? $bits['scheme'].':' : '')
	332	+ .(isset($bits['host']) ? '//'.$credentials.$bits['host'] : '')
	333	+ .(isset($bits['path']) ? $bits['path'] : '')
	334	+ .(!is_null($newQuery) ? '?'.$newQuery : '')
	335	+ .(isset($bits['fragment']) ? '#'.$bits['fragment'] : '')
	336	+ );
	337	+ $href = array_merge($href, $newUrl);
	338	+ endforeach;
	339	+ return array_unique($href);
	340	+ } /* FeedFinder::_url_manipulation_feeds () */
	341	+
174	342	function _tags ($tag) {
175	343	$html = $this->data();
176	344
177	345	// search through the HTML, save all <link> tags
		@@ -191,7 +359,7 @@
191	359	} /* foreach */
192	360	$ret[$n] = $final_link;
193	361	} /* for */
194	362	return $ret;
195		- }
	363	+ } /* FeedFinder::_tags () */
196	364	} /* class FeedFinder */
197	365