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 (intval($notice->is_local) === Notice::LOCAL_PUBLIC) {
  64             // Try to avoid actually mucking with the
  65             // notice content
  66             $c = $notice->content;
  67             $this->notice = $notice;
  68
  69             if(!$notice->getProfile()->
  70                 getPref("linkbackplugin", "disable_linkbacks")
  71             ) {
  72                 // Ignoring results
  73                 common_replace_urls_callback($c,
  74                                              array($this, 'linkbackUrl'));
  75             }
  76
  77             if($notice->isRepeat()) {
  78                 $repeat = Notice::getByID($notice->repeat_of);
  79                 $this->linkbackUrl($repeat->getUrl());
  80             } else if(!empty($notice->reply_to)) {
  81                 $parent = $notice->getParent();
  82                 $this->linkbackUrl($parent->getUrl());
  83             }
  84
  85             $replyProfiles = Profile::multiGet('id', $notice->getReplies());
  86             foreach($replyProfiles->fetchAll('profileurl') as $profileurl) {
  87                 $this->linkbackUrl($profileurl);
  88             }
  89         }
  90         return true;
  91     }
  92
  93     function linkbackUrl($url)
  94     {
  95         common_log(LOG_DEBUG,"Attempting linkback for " . $url);
  96
  97         $orig = $url;
  98         $url = htmlspecialchars_decode($orig);
  99         $scheme = parse_url($url, PHP_URL_SCHEME);
 100         if (!in_array($scheme, array('http', 'https'))) {
 101             return $orig;
 102         }
 103
 104         // XXX: Do a HEAD first to save some time/bandwidth
 105
 106         $fetcher = Auth_Yadis_Yadis::getHTTPFetcher();
 107
 108         $result = $fetcher->get($url,
 109                                 array('User-Agent: ' . $this->userAgent(),
 110                                       'Accept: application/html+xml,text/html'));
 111
 112         if (!in_array($result->status, array('200', '206'))) {
 113             return $orig;
 114         }
 115
 116         // XXX: Should handle relative-URI resolution in these detections
 117
 118         $wm = $this->getWebmention($result);
 119         if(!empty($wm)) {
 120             // It is the webmention receiver's job to resolve source
 121             // Ref: https://github.com/converspace/webmention/issues/43
 122             $this->webmention($url, $wm);
 123         } else {
 124             $pb = $this->getPingback($result);
 125             if (!empty($pb)) {
 126                 // Pingback still looks for exact URL in our source, so we
 127                 // must send what we have
 128                 $this->pingback($url, $pb);
 129             } else {
 130                 $tb = $this->getTrackback($result);
 131                 if (!empty($tb)) {
 132                     $this->trackback($result->final_url, $tb);
 133                 }
 134             }
 135         }
 136
 137         return $orig;
 138     }
 139
 140     // Based on https://github.com/indieweb/mention-client-php
 141     // which is licensed Apache 2.0
 142     function getWebmention($result) {
 143         // XXX: the fetcher only gives back one of each header, so this may fail on multiple Link headers
 144         if(preg_match('~<((?:https?://)?[^>]+)>; rel="webmention"~', $result->headers['Link'], $match)) {
 145             return $match[1];
 146         } elseif(preg_match('~<((?:https?://)?[^>]+)>; rel="http://webmention.org/?"~', $result->headers['Link'], $match)) {
 147             return $match[1];
 148         }
 149
 150         if(preg_match('/<(?:link|a)[ ]+href="([^"]+)"[ ]+rel="[^" ]* ?webmention ?[^" ]*"[ ]*\/?>/i', $result->body, $match)
 151            || preg_match('/<(?:link|a)[ ]+rel="[^" ]* ?webmention ?[^" ]*"[ ]+href="([^"]+)"[ ]*\/?>/i', $result->body, $match)) {
 152             return $match[1];
 153         } elseif(preg_match('/<(?:link|a)[ ]+href="([^"]+)"[ ]+rel="http:\/\/webmention\.org\/?"[ ]*\/?>/i', $result->body, $match)
 154                  || preg_match('/<(?:link|a)[ ]+rel="http:\/\/webmention\.org\/?"[ ]+href="([^"]+)"[ ]*\/?>/i', $result->body, $match)) {
 155             return $match[1];
 156         }
 157     }
 158
 159     function webmention($url, $endpoint) {
 160         $source = $this->notice->getUrl();
 161
 162         $payload = array(
 163             'source' => $source,
 164             'target' => $url
 165         );
 166
 167         $request = HTTPClient::start();
 168         try {
 169             $response = $request->post($endpoint,
 170                 array(
 171                     'Content-type: application/x-www-form-urlencoded',
 172                     'Accept: application/json'
 173                 ),
 174                 $payload
 175             );
 176
 177             if(!in_array($response->getStatus(), array(200,202))) {
 178                 common_log(LOG_WARNING,
 179                            "Webmention request failed for '$url' ($endpoint)");
 180             }
 181         } catch (HTTP_Request2_Exception $e) {
 182             common_log(LOG_WARNING,
 183                        "Webmention request failed for '$url' ($endpoint)");
 184         }
 185     }
 186
 187     function getPingback($result) {
 188         if (array_key_exists('X-Pingback', $result->headers)) {
 189             return $result->headers['X-Pingback'];
 190         } else if(preg_match('/<(?:link|a)[ ]+href="([^"]+)"[ ]+rel="[^" ]* ?pingback ?[^" ]*"[ ]*\/?>/i', $result->body, $match)
 191                   || preg_match('/<(?:link|a)[ ]+rel="[^" ]* ?pingback ?[^" ]*"[ ]+href="([^"]+)"[ ]*\/?>/i', $result->body, $match)) {
 192             return $match[1];
 193         }
 194     }
 195
 196     function pingback($url, $endpoint)
 197     {
 198         $args = array($this->notice->getUrl(), $url);
 199
 200         if (!extension_loaded('xmlrpc')) {
 201             if (!dl('xmlrpc.so')) {
 202                 common_log(LOG_ERR, "Can't pingback; xmlrpc extension not available.");
 203                 return;
 204             }
 205         }
 206
 207         $request = HTTPClient::start();
 208         try {
 209             $request->setBody(xmlrpc_encode_request('pingback.ping', $args));
 210             $response = $request->post($endpoint,
 211                 array('Content-Type: text/xml'),
 212                 false);
 213             $response = xmlrpc_decode($response->getBody());
 214             if (xmlrpc_is_fault($response)) {
 215                 common_log(LOG_WARNING,
 216                        "Pingback error for '$url' ($endpoint): ".
 217                        "$response[faultString] ($response[faultCode])");
 218             } else {
 219                 common_log(LOG_INFO,
 220                        "Pingback success for '$url' ($endpoint): ".
 221                        "'$response'");
 222             }
 223         } catch (HTTP_Request2_Exception $e) {
 224             common_log(LOG_WARNING,
 225                    "Pingback request failed for '$url' ($endpoint)");
 226         }
 227     }
 228
 229     // Largely cadged from trackback_cls.php by
 230     // Ran Aroussi <ran@blogish.org>, GPL2 or any later version
 231     // http://phptrackback.sourceforge.net/
 232     function getTrackback($result)
 233     {
 234         $text = $result->body;
 235         $url = $result->final_url;
 236
 237         if (preg_match_all('/(<rdf:RDF.*?<\/rdf:RDF>)/sm', $text, $match, PREG_SET_ORDER)) {
 238             for ($i = 0; $i < count($match); $i++) {
 239                 if (preg_match('|dc:identifier="' . preg_quote($url) . '"|ms', $match[$i][1])) {
 240                     $rdf_array[] = trim($match[$i][1]);
 241                 }
 242             }
 243
 244             // Loop through the RDFs array and extract trackback URIs
 245
 246             $tb_array = array(); // <- holds list of trackback URIs
 247
 248             if (!empty($rdf_array)) {
 249
 250                 for ($i = 0; $i < count($rdf_array); $i++) {
 251                     if (preg_match('/trackback:ping="([^"]+)"/', $rdf_array[$i], $array)) {
 252                         $tb_array[] = trim($array[1]);
 253                         break;
 254                     }
 255                 }
 256             }
 257
 258             // Return Trackbacks
 259
 260             if (empty($tb_array)) {
 261                 return null;
 262             } else {
 263                 return $tb_array[0];
 264             }
 265         }
 266
 267         if (preg_match_all('/(<a[^>]*?rel=[\'"]trackback[\'"][^>]*?>)/', $text, $match)) {
 268             foreach ($match[1] as $atag) {
 269                 if (preg_match('/href=[\'"]([^\'"]*?)[\'"]/', $atag, $url)) {
 270                     return $url[1];
 271                 }
 272             }
 273         }
 274
 275         return null;
 276
 277     }
 278
 279     function trackback($url, $endpoint)
 280     {
 281         $profile = $this->notice->getProfile();
 282
 283         // TRANS: Trackback title.
 284         // TRANS: %1$s is a profile nickname, %2$s is a timestamp.
 285         $args = array('title' => sprintf(_m('%1$s\'s status on %2$s'),
 286                                          $profile->nickname,
 287                                          common_exact_date($this->notice->created)),
 288                       'excerpt' => $this->notice->content,
 289                       'url' => $this->notice->getUrl(),
 290                       'blog_name' => $profile->nickname);
 291
 292         $fetcher = Auth_Yadis_Yadis::getHTTPFetcher();
 293
 294         $result = $fetcher->post($endpoint,
 295                                  http_build_query($args),
 296                                  array('User-Agent: ' . $this->userAgent()));
 297
 298         if ($result->status != '200') {
 299             common_log(LOG_WARNING,
 300                        "Trackback error for '$url' ($endpoint): ".
 301                        "$result->body");
 302         } else {
 303             common_log(LOG_INFO,
 304                        "Trackback success for '$url' ($endpoint): ".
 305                        "'$result->body'");
 306         }
 307     }
 308
 309     public function version()
 310     {
 311         return LINKBACKPLUGIN_VERSION;
 312     }
 313
 314     function onPluginVersion(array &$versions)
 315     {
 316         $versions[] = array('name' => 'Linkback',
 317                             'version' => LINKBACKPLUGIN_VERSION,
 318                             'author' => 'Evan Prodromou',
 319                             'homepage' => 'http://status.net/wiki/Plugin:Linkback',
 320                             'rawdescription' =>
 321                             // TRANS: Plugin description.
 322                             _m('Notify blog authors when their posts have been linked in '.
 323                                'microblog notices using '.
 324                                '<a href="http://www.hixie.ch/specs/pingback/pingback">Pingback</a> '.
 325                                'or <a href="http://www.movabletype.org/docs/mttrackback.html">Trackback</a> protocols.'));
 326         return true;
 327     }
 328
 329     public function onStartInitializeRouter(URLMapper $m)
 330     {
 331         $m->connect('settings/linkback', array('action' => 'linkbacksettings'));
 332         return true;
 333     }
 334
 335     function onEndAccountSettingsNav($action)
 336     {
 337         $action_name = $action->trimmed('action');
 338
 339         $action->menuItem(common_local_url('linkbacksettings'),
 340                           // TRANS: OpenID plugin menu item on user settings page.
 341                           _m('MENU', 'Send Linkbacks'),
 342                           // TRANS: OpenID plugin tooltip for user settings menu item.
 343                           _m('Opt-out of sending linkbacks.'),
 344                           $action_name === 'linkbacksettings');
 345         return true;
 346     }
 347 }