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