plugins/Linkback/LinkbackPlugin.php

   1 <?php
   2 /**
   3  * StatusNet, the distributed open-source microblogging tool
   4  *
   5  * Plugin to do linkbacks for notices containing links
   6  *
   7  * PHP version 5
   8  *
   9  * LICENCE: This program is free software: you can redistribute it and/or modify
  10  * it under the terms of the GNU Affero General Public License as published by
  11  * the Free Software Foundation, either version 3 of the License, or
  12  * (at your option) any later version.
  13  *
  14  * This program is distributed in the hope that it will be useful,
  15  * but WITHOUT ANY WARRANTY; without even the implied warranty of
  16  * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the
  17  * GNU Affero General Public License for more details.
  18  *
  19  * You should have received a copy of the GNU Affero General Public License
  20  * along with this program.  If not, see <http://www.gnu.org/licenses/>.
  21  *
  22  * @category  Plugin
  23  * @package   StatusNet
  24  * @author    Evan Prodromou <evan@status.net>
  25  * @copyright 2009 StatusNet, Inc.
  26  * @license   http://www.fsf.org/licensing/licenses/agpl-3.0.html GNU Affero General Public License version 3.0
  27  * @link      http://status.net/
  28  */
  29
  30 if (!defined('STATUSNET')) {
  31     exit(1);
  32 }
  33
  34 require_once('Auth/Yadis/Yadis.php');
  35
  36 define('LINKBACKPLUGIN_VERSION', '0.1');
  37
  38 /**
  39  * Plugin to do linkbacks for notices containing URLs
  40  *
  41  * After new notices are saved, we check their text for URLs. If there
  42  * are URLs, we test each URL to see if it supports any
  43  *
  44  * @category Plugin
  45  * @package  StatusNet
  46  * @author   Evan Prodromou <evan@status.net>
  47  * @license  http://www.fsf.org/licensing/licenses/agpl-3.0.html GNU Affero General Public License version 3.0
  48  * @link     http://status.net/
  49  *
  50  * @see      Event
  51  */
  52 class LinkbackPlugin extends Plugin
  53 {
  54     var $notice = null;
  55
  56     function __construct()
  57     {
  58         parent::__construct();
  59     }
  60
  61     function onHandleQueuedNotice($notice)
  62     {
  63         if ($notice->is_local == 1) {
  64             // Try to avoid actually mucking with the
  65             // notice content
  66             $c = $notice->content;
  67             $this->notice = $notice;
  68             // Ignoring results
  69             common_replace_urls_callback($c,
  70                                          array($this, 'linkbackUrl'));
  71
  72             if($notice->isRepeat()) {
  73                 $repeat = Notice::getByID($notice->repeat_of);
  74                 $this->linkbackUrl($repeat->getUrl());
  75             } else if(!empty($notice->reply_to)) {
  76                 $parent = $notice->getParent();
  77                 $this->linkbackUrl($parent->getUrl());
  78             }
  79         }
  80         return true;
  81     }
  82
  83     function linkbackUrl($url)
  84     {
  85         common_log(LOG_DEBUG,"Attempting linkback for " . $url);
  86
  87         $orig = $url;
  88         $url = htmlspecialchars_decode($orig);
  89         $scheme = parse_url($url, PHP_URL_SCHEME);
  90         if (!in_array($scheme, array('http', 'https'))) {
  91             return $orig;
  92         }
  93
  94         // XXX: Do a HEAD first to save some time/bandwidth
  95
  96         $fetcher = Auth_Yadis_Yadis::getHTTPFetcher();
  97
  98         $result = $fetcher->get($url,
  99                                 array('User-Agent: ' . $this->userAgent(),
 100                                       'Accept: application/html+xml,text/html'));
 101
 102         if (!in_array($result->status, array('200', '206'))) {
 103             return $orig;
 104         }
 105
 106         // XXX: Should handle relative-URI resolution in these detections
 107
 108         $wm = $this->getWebmention($result);
 109         if(!empty($wm)) {
 110             // It is the webmention receiver's job to resolve source
 111             // Ref: https://github.com/converspace/webmention/issues/43
 112             $this->webmention($url, $wm);
 113         } else {
 114             $pb = $this->getPingback($result);
 115             if (!empty($pb)) {
 116                 // Pingback still looks for exact URL in our source, so we
 117                 // must send what we have
 118                 $this->pingback($url, $pb);
 119             } else {
 120                 $tb = $this->getTrackback($result);
 121                 if (!empty($tb)) {
 122                     $this->trackback($result->final_url, $tb);
 123                 }
 124             }
 125         }
 126
 127         return $orig;
 128     }
 129
 130     // Based on https://github.com/indieweb/mention-client-php
 131     // which is licensed Apache 2.0
 132     function getWebmention($result) {
 133         // XXX: the fetcher only gives back one of each header, so this may fail on multiple Link headers
 134         if(preg_match('~<((?:https?://)?[^>]+)>; rel="webmention"~', $result->headers['Link'], $match)) {
 135             return $match[1];
 136         } elseif(preg_match('~<((?:https?://)?[^>]+)>; rel="http://webmention.org/?"~', $result->headers['Link'], $match)) {
 137             return $match[1];
 138         }
 139
 140         if(preg_match('/<(?:link|a)[ ]+href="([^"]+)"[ ]+rel="[^" ]* ?webmention ?[^" ]*"[ ]*\/?>/i', $result->body, $match)
 141            || preg_match('/<(?:link|a)[ ]+rel="[^" ]* ?webmention ?[^" ]*"[ ]+href="([^"]+)"[ ]*\/?>/i', $result->body, $match)) {
 142             return $match[1];
 143         } elseif(preg_match('/<(?:link|a)[ ]+href="([^"]+)"[ ]+rel="http:\/\/webmention\.org\/?"[ ]*\/?>/i', $result->body, $match)
 144                  || preg_match('/<(?:link|a)[ ]+rel="http:\/\/webmention\.org\/?"[ ]+href="([^"]+)"[ ]*\/?>/i', $result->body, $match)) {
 145             return $match[1];
 146         }
 147     }
 148
 149     function webmention($url, $endpoint) {
 150         $source = $this->notice->getUrl();
 151
 152         $payload = array(
 153             'source' => $source,
 154             'target' => $url
 155         );
 156
 157         $request = HTTPClient::start();
 158         try {
 159             $response = $request->post($endpoint,
 160                 array(
 161                     'Content-type: application/x-www-form-urlencoded',
 162                     'Accept: application/json'
 163                 ),
 164                 $payload
 165             );
 166
 167             if(!in_array($response->getStatus(), array(200,202))) {
 168                 common_log(LOG_WARNING,
 169                            "Webmention request failed for '$url' ($endpoint)");
 170             }
 171         } catch (HTTP_Request2_Exception $e) {
 172             common_log(LOG_WARNING,
 173                        "Webmention request failed for '$url' ($endpoint)");
 174         }
 175     }
 176
 177     function getPingback($result) {
 178         if (array_key_exists('X-Pingback', $result->headers)) {
 179             return $result->headers['X-Pingback'];
 180         } else if(preg_match('/<(?:link|a)[ ]+href="([^"]+)"[ ]+rel="[^" ]* ?pingback ?[^" ]*"[ ]*\/?>/i', $result->body, $match)
 181                   || preg_match('/<(?:link|a)[ ]+rel="[^" ]* ?pingback ?[^" ]*"[ ]+href="([^"]+)"[ ]*\/?>/i', $result->body, $match)) {
 182             return $match[1];
 183         }
 184     }
 185
 186     function pingback($url, $endpoint)
 187     {
 188         $args = array($this->notice->getUrl(), $url);
 189
 190         if (!extension_loaded('xmlrpc')) {
 191             if (!dl('xmlrpc.so')) {
 192                 common_log(LOG_ERR, "Can't pingback; xmlrpc extension not available.");
 193                 return;
 194             }
 195         }
 196
 197         $request = HTTPClient::start();
 198         try {
 199             $request->setBody(xmlrpc_encode_request('pingback.ping', $args));
 200             $response = $request->post($endpoint,
 201                 array('Content-Type: text/xml'),
 202                 false);
 203             $response = xmlrpc_decode($response->getBody());
 204             if (xmlrpc_is_fault($response)) {
 205                 common_log(LOG_WARNING,
 206                        "Pingback error for '$url' ($endpoint): ".
 207                        "$response[faultString] ($response[faultCode])");
 208             } else {
 209                 common_log(LOG_INFO,
 210                        "Pingback success for '$url' ($endpoint): ".
 211                        "'$response'");
 212             }
 213         } catch (HTTP_Request2_Exception $e) {
 214             common_log(LOG_WARNING,
 215                    "Pingback request failed for '$url' ($endpoint)");
 216         }
 217     }
 218
 219     // Largely cadged from trackback_cls.php by
 220     // Ran Aroussi <ran@blogish.org>, GPL2 or any later version
 221     // http://phptrackback.sourceforge.net/
 222     function getTrackback($result)
 223     {
 224         $text = $result->body;
 225         $url = $result->final_url;
 226
 227         if (preg_match_all('/(<rdf:RDF.*?<\/rdf:RDF>)/sm', $text, $match, PREG_SET_ORDER)) {
 228             for ($i = 0; $i < count($match); $i++) {
 229                 if (preg_match('|dc:identifier="' . preg_quote($url) . '"|ms', $match[$i][1])) {
 230                     $rdf_array[] = trim($match[$i][1]);
 231                 }
 232             }
 233
 234             // Loop through the RDFs array and extract trackback URIs
 235
 236             $tb_array = array(); // <- holds list of trackback URIs
 237
 238             if (!empty($rdf_array)) {
 239
 240                 for ($i = 0; $i < count($rdf_array); $i++) {
 241                     if (preg_match('/trackback:ping="([^"]+)"/', $rdf_array[$i], $array)) {
 242                         $tb_array[] = trim($array[1]);
 243                         break;
 244                     }
 245                 }
 246             }
 247
 248             // Return Trackbacks
 249
 250             if (empty($tb_array)) {
 251                 return null;
 252             } else {
 253                 return $tb_array[0];
 254             }
 255         }
 256
 257         if (preg_match_all('/(<a[^>]*?rel=[\'"]trackback[\'"][^>]*?>)/', $text, $match)) {
 258             foreach ($match[1] as $atag) {
 259                 if (preg_match('/href=[\'"]([^\'"]*?)[\'"]/', $atag, $url)) {
 260                     return $url[1];
 261                 }
 262             }
 263         }
 264
 265         return null;
 266
 267     }
 268
 269     function trackback($url, $endpoint)
 270     {
 271         $profile = $this->notice->getProfile();
 272
 273         // TRANS: Trackback title.
 274         // TRANS: %1$s is a profile nickname, %2$s is a timestamp.
 275         $args = array('title' => sprintf(_m('%1$s\'s status on %2$s'),
 276                                          $profile->nickname,
 277                                          common_exact_date($this->notice->created)),
 278                       'excerpt' => $this->notice->content,
 279                       'url' => $this->notice->getUrl(),
 280                       'blog_name' => $profile->nickname);
 281
 282         $fetcher = Auth_Yadis_Yadis::getHTTPFetcher();
 283
 284         $result = $fetcher->post($endpoint,
 285                                  http_build_query($args),
 286                                  array('User-Agent: ' . $this->userAgent()));
 287
 288         if ($result->status != '200') {
 289             common_log(LOG_WARNING,
 290                        "Trackback error for '$url' ($endpoint): ".
 291                        "$result->body");
 292         } else {
 293             common_log(LOG_INFO,
 294                        "Trackback success for '$url' ($endpoint): ".
 295                        "'$result->body'");
 296         }
 297     }
 298
 299     public function version()
 300     {
 301         return LINKBACKPLUGIN_VERSION;
 302     }
 303
 304     function onPluginVersion(array &$versions)
 305     {
 306         $versions[] = array('name' => 'Linkback',
 307                             'version' => LINKBACKPLUGIN_VERSION,
 308                             'author' => 'Evan Prodromou',
 309                             'homepage' => 'http://status.net/wiki/Plugin:Linkback',
 310                             'rawdescription' =>
 311                             // TRANS: Plugin description.
 312                             _m('Notify blog authors when their posts have been linked in '.
 313                                'microblog notices using '.
 314                                '<a href="http://www.hixie.ch/specs/pingback/pingback">Pingback</a> '.
 315                                'or <a href="http://www.movabletype.org/docs/mttrackback.html">Trackback</a> protocols.'));
 316         return true;
 317     }
 318 }