← All changes | feedfinder.class.php +156 -41 2010.0528 → 2011.0211 View file →

		@@ -1,15 +1,23 @@
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;
8	15	var $_cache_uri = NULL;
9	16
10	17	var $verify = FALSE;
11		-
	18	+ var $fallbacks = 3;
	19	+
12	20	var $_response = NULL;
13	21	var $_data = NULL;
14	22	var $_error = NULL;
15	23	var $_head = NULL;
		@@ -23,56 +31,81 @@
23	31	'application/x-atom+xml'
24	32	);
25	33	var $_feed_markers = array('\\<feed', '\\<rss', 'xmlns="http://purl.org/rss/1.0');
26	34	var $_html_markers = array('\\<html');
	35	+ var $_opml_markers = array('\\<opml', '\\<outline');
27	36	var $_obvious_feed_url = array('[./]rss', '[./]rdf', '[./]atom', '[./]feed', '\.xml');
28	37	var $_maybe_feed_url = array ('rss', 'rdf', 'atom', 'feed', 'xml');
29	38
30		- function FeedFinder ($uri = NULL, $verify = TRUE) {
	39	+ function FeedFinder ($uri = NULL, $verify = TRUE, $fallbacks = 3) {
31	40	$this->uri = $uri; $this->verify = $verify;
	41	+ $this->fallbacks = $fallbacks;
32	42	} /* FeedFinder::FeedFinder () */
33	43
34	44	function find ($uri = NULL) {
35	45	$ret = array ();
36		- if (!is_null($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 = $this->_a_href_feeds(FALSE);
	46	+ if (!is_null($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;
49	73
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);
	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);
53	81
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);
	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);
68	96	} /* FeedFinder::find () */
69	97
70	98	function data ($uri = NULL) {
71	99	$this->_get($uri);
72	100	return $this->_data;
73		- }
	101	+ } /* FeedFinder::data () */
74	102
	103	+ function upload_data ($data) {
	104	+ $this->uri = 'tag:localhost';
	105	+ $this->_data = $data;
	106	+ } /* FeedFinder::upload_data () */
	107	+
75	108	function status ($uri = NULL) {
76	109	$this->_get($uri);
77	110
78	111	if (!is_wp_error($this->_response) and isset($this->_response['response']['code'])) :
		@@ -108,14 +141,24 @@
108	141	)
109	142	);
110	143	} /* FeedFinder::is_feed () */
111	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	+ $data
	151	+ )
	152	+ );
	153	+ } /* FeedFinder::is_opml () */
	154	+
112	155	# --- Private methods ---
113	156	function _get ($uri = NULL) {
114	157	if ($uri) $this->uri = $uri;
115	158
116	159	// Is the result not yet cached?
117		- if ($this->_cache_uri !== $this->uri) :
	160	+ if ($this->uri != 'tag:localhost' and $this->_cache_uri !== $this->uri) :
118	161	$headers['Connection'] = 'close';
119	162	$headers['Accept'] = 'application/atom+xml application/rdf+xml application/rss+xml application/xml text/html */*';
120	163	$headers['User-Agent'] = 'feedfinder/1.2 (compatible; PHP FeedFinder) +http://projects.radgeek.com/feedwordpress';
121	164
		@@ -121,9 +164,9 @@
121	164
122	165	// Use WordPress API function
123	166	$client = wp_remote_request($this->uri, array(
124	167	'headers' => $headers,
125		- 'timeout' => FEEDWORDPRESS_FETCH_TIME_OUT,
	168	+ 'timeout' => FeedWordPress::fetch_timeout(),
126	169	));
127	170
128	171	$this->_response = $client;
129	172	if (is_wp_error($client)) :
		@@ -138,8 +181,25 @@
138	181	$this->_cache_uri = $this->uri;
139	182	endif;
140	183	} /* FeedFinder::_get () */
141	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	+
142	202	function _link_rel_feeds () {
143	203	$links = $this->_tags('link');
144	204	$link_count = count($links);
145	205
		@@ -152,9 +212,9 @@
152	212	} /* if */
153	213	} /* if */
154	214	} /* for */
155	215	return $href;
156		- }
	216	+ } /* FeedFinder::_link_rel_feeds () */
157	217
158	218	function _a_href_feeds ($obvious = TRUE) {
159	219	$pattern = ($obvious ? $this->_obvious_feed_url : $this->_maybe_feed_url);
160	220
		@@ -168,10 +228,65 @@
168	228	$href[] = $links[$n]['href'];
169	229	} /* if */
170	230	} /* for */
171	231	return $href;
172		- }
	232	+ } /* FeedFinder::_link_a_href_feeds () */
173	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	+
174	289	function _tags ($tag) {
175	290	$html = $this->data();
176	291
177	292	// search through the HTML, save all <link> tags
		@@ -191,7 +306,7 @@
191	306	} /* foreach */
192	307	$ret[$n] = $final_link;
193	308	} /* for */
194	309	return $ret;
195		- }
	310	+ } /* FeedFinder::_tags () */
196	311	} /* class FeedFinder */
197	312