← All changes | feedfinder.class.php +215 -43 2010.0528 → 2017.1004 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 +32,130 @@
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 __construct( $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 FeedFinder( $uri = NULL, $params = array(), $fallbacks = 3 ) {
	63	+ self::__construct( $uri, $params, $fallbacks );
	64	+ }
	65	+
	66	+ function find ($uri = NULL, $params = array()) {
	67	+ $params = wp_parse_args($params, array( // Defaults
	68	+ "authentication" => -1,
	69	+ "username" => NULL,
	70	+ "password" => NULL,
	71	+ ));
	72	+
	73	+ // Equivalents
	74	+ if ($params['authentication']=='-') :
	75	+ $params['authentication'] = NULL;
	76	+ $params['username'] = NULL;
	77	+ $params['password'] = NULL;
	78	+ endif;
	79	+
	80	+ // Set/reset
	81	+ if ($params['authentication'] != -1) :
	82	+ $this->credentials = array(
	83	+ "authentication" => $params['authentication'],
	84	+ "username" => $params['username'],
	85	+ "password" => $params['password'],
	86	+ );
	87	+ endif;
	88	+
35	89	$ret = array ();
36		- if (!is_null($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 = $this->_a_href_feeds(FALSE);
	90	+ if (!is_null($this->data($uri))) :
	91	+ if ($this->is_opml($uri)) :
	92	+ $href = $this->_opml_rss_uris();
	93	+ else :
	94	+ if ($this->is_feed($uri)) :
	95	+ $href = array($this->uri);
	96	+ else :
	97	+ // Assume that we have HTML or XHTML (even if we don't, who's
	98	+ // it gonna hurt?) Autodiscovery is the preferred method.
	99	+ $href = $this->_link_rel_feeds();
	100	+
	101	+ // ... but we'll also take the little orange buttons
	102	+ if ($this->fallbacks > 0) :
	103	+ $href = array_merge($href, $this->_a_href_feeds(TRUE));
	104	+ endif;
	105	+
	106	+ // If all that failed, look harder
	107	+ if ($this->fallbacks > 1) :
	108	+ if (count($href) == 0) :
	109	+ $href = $this->_a_href_feeds(FALSE);
	110	+ endif;
	111	+ endif;
	112	+
	113	+ // Our search may turn up duplicate URIs. We only need to do
	114	+ // any given URI once. Props to Camilo <http://projects.radgeek.com/2008/12/14/feedwordpress-20081214/#comment-20090122160414>
	115	+ $href = array_unique($href);
	116	+ endif;
49	117
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);
	118	+ // Try some clever URL little tricks before we go
	119	+ if ($this->fallbacks > 2) :
	120	+ $href = array_merge($href, $this->_url_manipulation_feeds());
	121	+ endif;
	122	+ endif;
	123	+
	124	+ $href = array_unique($href);
53	125
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);
	126	+ // Verify feeds and resolve relative URIs
	127	+ foreach ($href as $u) :
	128	+ $the_uri = SimplePie_Misc::absolutize_url($u, $this->uri);
	129	+ if ($this->verify and ($u != $this->uri and $the_uri != $this->uri)) :
	130	+ $feed = new FeedFinder($the_uri, $this->credentials);
	131	+ if ($feed->is_feed()) : $ret[] = $the_uri; endif;
	132	+ unset($feed);
	133	+ else :
	134	+ $ret[] = $the_uri;
	135	+ endif;
	136	+ endforeach;
	137	+ endif;
	138	+
	139	+ if ($this->is_401($uri)) :
	140	+ $ret = array_merge(array(
	141	+ new WP_Error('http_request_failed', '401 Not authorized', array("uri" => $this->uri, "status" => 401)),
	142	+ ), $ret);
	143	+ endif;
	144	+
	145	+ return array_values($ret);
68	146	} /* FeedFinder::find () */
69	147
70	148	function data ($uri = NULL) {
71	149	$this->_get($uri);
72	150	return $this->_data;
73		- }
	151	+ } /* FeedFinder::data () */
74	152
	153	+ function upload_data ($data) {
	154	+ $this->uri = 'tag:localhost';
	155	+ $this->_data = $data;
	156	+ } /* FeedFinder::upload_data () */
	157	+
75	158	function status ($uri = NULL) {
76	159	$this->_get($uri);
77	160
78	161	if (!is_wp_error($this->_response) and isset($this->_response['response']['code'])) :
		@@ -94,8 +177,12 @@
94	177	endif;
95	178	return $message;
96	179	}
97	180
	181	+ function is_401 ($uri = NULL) {
	182	+ return (intval($this->status($uri))==401);
	183	+ } /* FeedFinder::is_401 () */
	184	+
98	185	function is_feed ($uri = NULL) {
99	186	$data = $this->data($uri);
100	187
101	188	return (
		@@ -108,22 +195,35 @@
108	195	)
109	196	);
110	197	} /* FeedFinder::is_feed () */
111	198
	199	+ function is_opml ($uri = NULL) {
	200	+ $data = $this->data($uri);
	201	+ return (
	202	+ preg_match (
	203	+ "\007(".implode('|',$this->_opml_markers).")\007i",
	204	+ $data
	205	+ )
	206	+ );
	207	+ } /* FeedFinder::is_opml () */
	208	+
112	209	# --- Private methods ---
113	210	function _get ($uri = NULL) {
114	211	if ($uri) $this->uri = $uri;
115	212
116	213	// Is the result not yet cached?
117		- if ($this->_cache_uri !== $this->uri) :
	214	+ if ($this->uri != 'tag:localhost' and $this->_cache_uri !== $this->uri) :
118	215	$headers['Connection'] = 'close';
119	216	$headers['Accept'] = 'application/atom+xml application/rdf+xml application/rss+xml application/xml text/html */*';
120	217	$headers['User-Agent'] = 'feedfinder/1.2 (compatible; PHP FeedFinder) +http://projects.radgeek.com/feedwordpress';
121	218
122	219	// Use WordPress API function
123		- $client = wp_remote_request($this->uri, array(
	220	+ $client = wp_remote_request($this->uri, array_merge(
	221	+ $this->credentials,
	222	+ array(
124	223	'headers' => $headers,
125		- 'timeout' => FEEDWORDPRESS_FETCH_TIME_OUT,
	224	+ 'timeout' => FeedWordPress::fetch_timeout(),
	225	+ )
126	226	));
127	227
128	228	$this->_response = $client;
129	229	if (is_wp_error($client)) :
		@@ -138,8 +238,25 @@
138	238	$this->_cache_uri = $this->uri;
139	239	endif;
140	240	} /* FeedFinder::_get () */
141	241
	242	+ function _opml_rss_uris () {
	243	+ // Really we should parse the XML and use the structure to
	244	+ // return something intelligent to programs that want to use it
	245	+ // Oh babe! maybe some day...
	246	+
	247	+ $opml = $this->data();
	248	+
	249	+ $rx = FeedWordPressHTML::attributeRegex('outline', 'xmlUrl');
	250	+ if (preg_match_all($rx, $opml, $matches, PREG_SET_ORDER)) :
	251	+ foreach ($matches as $m) :
	252	+ $match = FeedWordPressHTML::attributeMatch($m);
	253	+ $r[] = $match['value'];
	254	+ endforeach;
	255	+ endif;
	256	+ return $r;
	257	+ } /* FeedFinder::_opml_rss_uris () */
	258	+
142	259	function _link_rel_feeds () {
143	260	$links = $this->_tags('link');
144	261	$link_count = count($links);
145	262
		@@ -152,9 +269,9 @@
152	269	} /* if */
153	270	} /* if */
154	271	} /* for */
155	272	return $href;
156		- }
	273	+ } /* FeedFinder::_link_rel_feeds () */
157	274
158	275	function _a_href_feeds ($obvious = TRUE) {
159	276	$pattern = ($obvious ? $this->_obvious_feed_url : $this->_maybe_feed_url);
160	277
		@@ -168,10 +285,65 @@
168	285	$href[] = $links[$n]['href'];
169	286	} /* if */
170	287	} /* for */
171	288	return $href;
172		- }
	289	+ } /* FeedFinder::_link_a_href_feeds () */
173	290
	291	+ function _url_manipulation_feeds () {
	292	+ $href = array();
	293	+
	294	+ // check for HTTP GET parameters that look feed-like.
	295	+ $bits = parse_url($this->uri);
	296	+ foreach (array('rss', 'rss2', 'atom', 'rdf') as $format) :
	297	+ if (isset($bits['query']) and (strlen($bits['query']) > 0)) :
	298	+ $newQuery = preg_replace('/([?&=])(rss2?|atom|rdf)/i', '$1'.$format, $bits['query']);
	299	+ else :
	300	+ $newQuery = NULL;
	301	+ endif;
	302	+
	303	+ if (isset($bits['path']) and (strlen($bits['path']) > 0)) :
	304	+ $newPath = preg_replace('!([/.])(rss2?|atom|rdf)!i', '$1'.$format, $bits['path']);
	305	+ else :
	306	+ $newPath = NULL;
	307	+ endif;
	308	+
	309	+ // Reassemble, check and return
	310	+ $credentials = '';
	311	+ if (isset($bits['user'])) :
	312	+ $credentials = $bits['user'];
	313	+ if (isset($bits['pass'])) :
	314	+ $credentials .= ':'.$bits['pass'];
	315	+ endif;
	316	+ $credentials .= '@';
	317	+ endif;
	318	+
	319	+ // Variations on a theme
	320	+ $newUrl[0] = (''
	321	+ .(isset($bits['scheme']) ? $bits['scheme'].':' : '')
	322	+ .(isset($bits['host']) ? '//'.$credentials.$bits['host'] : '')
	323	+ .(!is_null($newPath) ? $newPath : '')
	324	+ .(!is_null($newQuery) ? '?'.$newQuery : '')
	325	+ .(isset($bits['fragment']) ? '#'.$bits['fragment'] : '')
	326	+ );
	327	+ $newUrl[1] = (''
	328	+ .(isset($bits['scheme']) ? $bits['scheme'].':' : '')
	329	+ .(isset($bits['host']) ? '//'.$credentials.$bits['host'] : '')
	330	+ .(!is_null($newPath) ? $newPath : '')
	331	+ .(isset($bits['query']) ? '?'.$bits['query'] : '')
	332	+ .(isset($bits['fragment']) ? '#'.$bits['fragment'] : '')
	333	+ );
	334	+ $newUrl[2] = (''
	335	+ .(isset($bits['scheme']) ? $bits['scheme'].':' : '')
	336	+ .(isset($bits['host']) ? '//'.$credentials.$bits['host'] : '')
	337	+ .(isset($bits['path']) ? $bits['path'] : '')
	338	+ .(!is_null($newQuery) ? '?'.$newQuery : '')
	339	+ .(isset($bits['fragment']) ? '#'.$bits['fragment'] : '')
	340	+ );
	341	+ $href = array_merge($href, $newUrl);
	342	+ endforeach;
	343	+ return array_unique($href);
	344	+ } /* FeedFinder::_url_manipulation_feeds () */
	345	+
174	346	function _tags ($tag) {
175	347	$html = $this->data();
176	348
177	349	// search through the HTML, save all <link> tags
		@@ -191,7 +363,7 @@
191	363	} /* foreach */
192	364	$ret[$n] = $final_link;
193	365	} /* for */
194	366	return $ret;
195		- }
	367	+ } /* FeedFinder::_tags () */
196	368	} /* class FeedFinder */
197	369