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