Context Navigation

TrackbackUtil.class.php @ 150

Revision 150, 12.6 kB (checked in by d0nut, 5 years ago)
added documentation added time to incoming trackbacks shortended rdf-description to 255 chars encoded incoming trackback-data to html entities (XSS secure) added icon for active trackbacks

Line
1	<?php
2	// import seorewriter
3	require_once(WCF_DIR.'lib/util/StringUtil.class.php');
4	require_once(WCF_DIR.'lib/page/PublicSEORewriter.class.php');
5	require_once(WCF_DIR.'lib/util/FileUtil.class.php');
6
7	/**
8	* trackback util
9	*
10	* @author Torben Brodt
11	* @package de.easy-coding.wbb.trackback
12	* @license GNU General Public License <http://opensource.org/licenses/gpl-3.0.html>
13	*/
14	class TrackbackUtil {
15	public static $agent = 'WBB Trackback Mod';
16	protected static $timeout_bytes = 51200;
17
18	/**
19	* fetchs necessery postdata
20	* @param postIDs int[] list of postids with possible trackback data
21	* @return mixed[] including all threaddata + threadurl
22	*/
23	public static function getPost($postIDs) {
24	if(!is_array($postIDs))
25	$postIDs = array($postIDs);
26
27	// fetch thread data
28	$sql = "SELECT thread.*
29	FROM wbb".WBB_N."_post post
30	LEFT JOIN wbb".WBB_N."_thread thread
31	ON (thread.threadID=post.threadID)
32	WHERE postID IN (".implode(',', $postIDs).") ";
33
34	$row = WCF::getDB()->getFirstRow($sql);
35
36	$rewriter = new PublicSEORewriter();
37
38	// public seo rewriter: cache
39	$rewriter->publicCacheThreads($row['threadID'], $row);
40
41	// public seo rewriter use
42	$row['url'] = StringUtil::decodeHTML(FileUtil::addTrailingSlash(PAGE_URL).$rewriter->publicParseThreadURLs($row['threadID'], ''));
43
44	return $row;
45	}
46
47	/**
48	* returns rdf string
49	* @param postPermalink string permalink
50	* @param postID int postid
51	* @param postTopic string posttopic
52	* @param postPreview string postpreview
53	* @param postUsername string postusername
54	* @param postTime int posttime as unix timestamp
55	*/
56	public static function getRDF($postPermalink, $postID, $postTopic, $postPreview, $postUsername, $postTime) {
57	$postPermalink = StringUtil::decodeHTML($postPermalink);
58	$page_url = FileUtil::addTrailingSlash(PAGE_URL);
59	$postPreview = str_replace(array("\n","\r"), ' ', $postPreview);
60	$postPreview = str_replace(array(" "," "," "), ' ', $postPreview);
61	$postPreview = substr($postPreview, 0, 255);
62	return '<link rel="pingback" href="'.$page_url.'index.php?action=Pingback" />
63	<!--
64	<rdf:RDF xmlns:rdf="http://www.w3.org/1999/02/22-rdf-syntax-ns#"
65	xmlns:dc="http://purl.org/dc/elements/1.1/"
66	xmlns:trackback="http://madskills.com/public/xml/rss/module/trackback/">
67	<rdf:Description
68	rdf:about="'.$postPermalink.'"
69	dc:identifier="'.$postPermalink.'"
70	trackback:ping="'.$page_url.'index.php?action=Trackback&postID='.$postID.'"
71	dc:title="'.$postTopic.'"
72	dc:subject="TrackBack"
73	dc:description="'.$postPreview.'"
74	dc:creator="'.$postUsername.'"
75	dc:date="'.date('r', $postTime).'" />
76	</rdf:RDF>
77	-->';
78	}
79
80	/**
81	* saves incoming ping/trackback
82	* @param postID int postid
83	* @param postURL string post-/thread-url
84	* @param title string title
85	* @param excerpt string excerpt
86	* @param url string url
87	* @param blog_name string blog_name
88	*/
89	public static function save($postID, $postURL, $title, $excerpt, $url, $blog_name) {
90	// akismet check spam
91	$isSpam = self::isSpam($title, $excerpt, $url, $blog_name);
92
93	// escape html characters
94	$postURL = StringUtil::encodeHTML($postURL);
95	$title = StringUtil::encodeHTML($title);
96	$excerpt = StringUtil::encodeHTML($excerpt);
97	$url = StringUtil::encodeHTML($url);
98	$blog_name = StringUtil::encodeHTML($blog_name);
99
100	// insert trackback data
101	$sql = "INSERT INTO wbb".WBB_N."_trackback
102	(
103	postID,
104	postURL,
105	title,
106	excerpt,
107	url,
108	blog_name,
109	time,
110	isSpam
111	) VALUES (
112	{$postID},
113	'".escapeString($postURL)."',
114	'".escapeString($title)."',
115	'".escapeString($excerpt)."',
116	'".escapeString($url)."',
117	'".escapeString($blog_name)."',
118	NOW(),
119	".intval($isSpam)."
120	);";
121
122	WBBCore::getDB()->sendQuery($sql);
123	}
124
125	/**
126	* does track and ping operations on alien urls
127	* @param postID int postid
128	* @param alienurl string url in content - ping and trackback here
129	* @param page_title string own forum title
130	* @param author string own author
131	* @param url string own url
132	* @param title string own post title
133	* @param excerpt string own excerpt
134	*/
135	public static function trackAndPing($postID, $alienurl, $page_title, $author, $url, $title, $excerpt) {
136	$content = ""; // hold as reference
137
138	// auto discovery
139	$pingbacks = self::discover_pingback_uri($alienurl, $content);
140	$trackbacks = count($pingbacks) > 0 ? array() : self::discover_trackback_uri($content, $alienurl);
141
142	// merge pingbacks with public pingback services
143	$pingbacks = array_merge($pingbacks, explode("\n", StringUtil::unifyNewlines(MESSAGE_PINGBACK)));
144	$pingbacks = array_filter($pingbacks, create_function('$a','return !empty($a);'));
145
146	// send pingbacks
147	foreach($pingbacks as $pingbackurl) {
148	if(self::send_pingback($alienurl, $pingbackurl, $url)) {
149	self::save_log($postID, $alienurl);
150	}
151	}
152
153	// send trackbacks
154	foreach($trackbacks as $trackbackurl) {
155	if(self::send_trackback($trackbackurl, $page_title, $author, $url, $title, $excerpt)) {
156	self::save_log($postID, $alienurl);
157	}
158	}
159	}
160
161	/**
162	* akismet check
163	* @param title string incoming title
164	* @param excerpt string incoming excerpt
165	* @param url string incoming url
166	* @param blog_name string incoming blog_name
167	* @return boolean true, if spam
168	*/
169	protected static function isSpam($title, $excerpt, $url, $blog_name) {
170	if(AKISMET_API_KEY == '')
171	return false;
172
173	// WCF includes
174	require_once(WCF_DIR.'lib/util/AkismetUtil.class.php');
175
176	$akismet = new AkismetUtil(AKISMET_API_KEY, PAGE_URL);
177	$akismet->setCommentType('trackback');
178	$akismet->setPermalink($myurl); //absolute
179
180	// assign post data
181	$akismet->setCommentAuthor($title);
182	$akismet->setCommentAuthorURL($url);
183	$akismet->setCommentContent($excerpt);
184
185	return $akismet->isCommentSpam();
186	}
187
188	/**
189	* discover pingback uri
190	* @param url string the alien url
191	* @param contents (reference) string empty variable - but the content of the alien url will be returned here
192	* @return pingbacks string[] the pingbacks of the alien url (should be just one)
193	*/
194	protected static function discover_pingback_uri($url, &$contents) {
195	$pingbacks = array(); // return var
196
197	$byte_count = 0;
198	$contents = '';
199	$headers = '';
200	$pingback_str_dquote = 'rel="pingback"';
201	$pingback_str_squote = 'rel=\'pingback\'';
202	$x_pingback_str = 'x-pingback: ';
203	$pingback_href_original_pos = 27;
204
205	extract(parse_url($url), EXTR_SKIP);
206
207	if ( !isset($host) ) // Not an URL. This should never happen.
208	return false;
209
210	$path = ( !isset($path) ) ? '/' : $path;
211	$path .= ( isset($query) ) ? '?' . $query : '';
212	$port = ( isset($port) ) ? $port : 80;
213
214	// Try to connect to the server at $host
215	$fp = @fsockopen($host, $port, $errno, $errstr, 2);
216	if ( !$fp ) // Couldn't open a connection to $host
217	return false;
218
219	// Send the GET request
220	$request = "GET $path HTTP/1.1\r\nHost: $host\r\nUser-Agent: ".self::$agent." \r\n\r\n";
221	// ob_end_flush();
222	fputs($fp, $request);
223
224	// Let's check for an X-Pingback header first
225	while ( !feof($fp) ) {
226	$line = fgets($fp, 512);
227	if ( trim($line) == '' )
228	break;
229	$headers .= trim($line)."\n";
230	$x_pingback_header_offset = strpos(strtolower($headers), $x_pingback_str);
231	if ( $x_pingback_header_offset ) {
232	// We got it!
233	preg_match('#x-pingback: (.+)#is', $headers, $matches);
234	$pingbacks[] = trim($matches[1]);
235	return $pingbacks;
236	}
237	if ( strpos(strtolower($headers), 'content-type: ') ) {
238	preg_match('#content-type: (.+)#is', $headers, $matches);
239	$content_type = trim($matches[1]);
240	}
241	}
242
243	if ( preg_match('#(image\|audio\|video\|model)/#is', $content_type) ) // Not an (x)html, sgml, or xml page, no use going further
244	return false;
245
246	while ( !feof($fp) ) {
247	$line = fgets($fp, 1024);
248	$contents .= trim($line);
249	$pingback_link_offset_dquote = strpos($contents, $pingback_str_dquote);
250	$pingback_link_offset_squote = strpos($contents, $pingback_str_squote);
251	if ( $pingback_link_offset_dquote \|\| $pingback_link_offset_squote ) {
252	$quote = ($pingback_link_offset_dquote) ? '"' : '\'';
253	$pingback_link_offset = ($quote=='"') ? $pingback_link_offset_dquote : $pingback_link_offset_squote;
254	$pingback_href_pos = @strpos($contents, 'href=', $pingback_link_offset);
255	$pingback_href_start = $pingback_href_pos+6;
256	$pingback_href_end = @strpos($contents, $quote, $pingback_href_start);
257	$pingback_server_url_len = $pingback_href_end - $pingback_href_start;
258	$pingback_server_url = substr($contents, $pingback_href_start, $pingback_server_url_len);
259	// We may find rel="pingback" but an incomplete pingback URL
260	if ( $pingback_server_url_len > 0 ) // We got it!
261	$pingbacks[] = $pingback_server_url;
262	break;
263	}
264	$byte_count += strlen($line);
265	if ( $byte_count > self::$timeout_bytes ) {
266	// It's no use going further, there probably isn't any pingback
267	// server to find in this file. (Prevents loading large files.)
268	break;
269	}
270	}
271
272	return $pingbacks;
273	}
274
275	/**
276	* discover trackback uri
277	* @param contents string html content to parse
278	* @param url string the url, which should allow trackback:ping
279	* @return trackbacks string[] the trackbacks of the alien url (should be just one)
280	*/
281	protected static function discover_trackback_uri($contents, $url) {
282	$rdf = array(); // <- holds list of RDF segments
283
284	if ($contents) {
285	preg_match_all('/(<rdf:RDF.*?<\/rdf:RDF>)/sm', $contents, $link_rdf, PREG_SET_ORDER);
286
287	// Loop through all rdf segments
288	foreach ($link_rdf as $seg) {
289	if (preg_match('\|dc:identifier="' . preg_quote($url) . '"\|ms', $seg[1])) {
290	$rdf[] = trim($seg[1]);
291	}
292	}
293	}
294	// Loop through the RDFs array and extract trackback URIs
295	$trackbacks = array(); // <- holds list of trackback URIs
296	foreach($rdf as $rdf_url) {
297	if (preg_match('/trackback:ping="([^"]+)"/', $rdf_url, $array)) {
298	$trackbacks[] = trim($array[1]);
299	}
300	}
301
302	return $trackbacks;
303	}
304
305
306	/**
307	* Send a Pingback
308	* @param alienurl string the url, we want to ping
309	* @param pingbackurl string the pingback url from the destination site (e.g. xmlrpc.php)
310	* @param url string our own url
311	* @return boolean true if success
312	*/
313	protected static function send_pingback($alienurl, $pingbackurl, $url) {
314	require_once(WBB_DIR.'lib/util/IXR.class.php');
315	extract(parse_url($pingbackurl), EXTR_SKIP);
316
317	if ( !isset($host) ) // Not an URL. This should never happen.
318	return false;
319
320	$path = ( !isset($path) ) ? '/' : $path;
321	$path .= ( isset($query) ) ? '?' . $query : '';
322	$port = ( isset($port) ) ? $port : 80;
323
324	// using a timeout of 3 seconds should be enough to cover slow servers
325	$client = new IXR_Client($host, $path);
326	$client->timeout = 3;
327	$client->useragent = self::$agent;
328
329	// when set to true, this outputs debug messages by itself
330	$client->debug = false;
331
332	// Already registered
333	return $client->query('pingback.ping', $url, $alienurl) \|\| (isset($client->error->code) && 48 == $client->error->code);
334	}
335
336	/**
337	* Send a Trackback
338	* @param trackbackurl string the url, we want to trackback
339	* @param page_title string own forum title
340	* @param author string own author (ignored)
341	* @param url string own url
342	* @param title string own post title
343	* @param excerpt string own excerpt
344	* @return resource pointer on socket
345	*/
346	protected static function send_trackback($trackbackurl, $page_title, $author, $url, $title, $excerpt) {
347	$blog_title = urlencode($page_title);
348	$url = urlencode($url);
349	$title = urlencode($title);
350	$excerpt = urlencode($excerpt);
351
352	$query_string = "title={$title}&url={$url}&blog_name={$blog_title}&excerpt={$excerpt}";
353
354	$parse = parse_url($trackbackurl);
355	$http_request = 'POST ' . $parse['path'] . (isset($parse['query']) ? '?'.$parse['query'] : '') . " HTTP/1.0\r\n";
356	$http_request .= 'Host: '.$parse['host']."\r\n";
357	$http_request .= 'Content-Type: application/x-www-form-urlencoded; charset='.CHARSET."\r\n";
358	$http_request .= 'Content-Length: '.strlen($query_string)."\r\n";
359	$http_request .= "User-Agent: {self::agent}";
360	$http_request .= "\r\n\r\n";
361	$http_request .= $query_string;
362
363	$fs = @fsockopen($parse['host'], isset($parse['port'])?$parse['port']:80, $errno, $errstr, 4);
364	@fputs($fs, $http_request);
365	@fclose($fs);
366
367	return $fs;
368	}
369
370	/**
371	* saves log
372	* @param postID int postID
373	* @param alienURL string alienURL
374	*/
375	protected static function save_log($postID, $alienurl) {
376	$sql = "INSERT INTO wbb".WBB_N."_trackbackLog
377	(
378	postID,
379	alienURL,
380	timestamp
381	) VALUES (
382	{$postID},
383	'".escapeString($alienurl)."',
384	".time()."
385	);";
386
387	@WBBCore::getDB()->sendQuery($sql); // ignore error messages...
388	}
389	}
390	?>

Note: See TracBrowser for help on using the browser.

Context Navigation

root/trackback/files/lib/util/TrackbackUtil.class.php @ 150

Download in other formats: