plugins/ActivitySpam/spamfilter.php

   1 <?php
   2   /**
   3    * StatusNet - the distributed open-source microblogging tool
   4    * Copyright (C) 2012, StatusNet, Inc.
   5    *
   6    * Spam filter class
   7    *
   8    * PHP version 5
   9    *
  10    * This program is free software: you can redistribute it and/or modify
  11    * it under the terms of the GNU Affero General Public License as published by
  12    * the Free Software Foundation, either version 3 of the License, or
  13    * (at your option) any later version.
  14    *
  15    * This program is distributed in the hope that it will be useful,
  16    * but WITHOUT ANY WARRANTY; without even the implied warranty of
  17    * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the
  18    * GNU Affero General Public License for more details.
  19    *
  20    * You should have received a copy of the GNU Affero General Public License
  21    * along with this program.  If not, see <http://www.gnu.org/licenses/>.
  22    *
  23    * @category  Spam
  24    * @package   StatusNet
  25    * @author    Evan Prodromou <evan@status.net>
  26    * @copyright 2012 StatusNet, Inc.
  27    * @license   http://www.fsf.org/licensing/licenses/agpl-3.0.html AGPL 3.0
  28    * @link      http://status.net/
  29    */
  30
  31 if (!defined('STATUSNET')) {
  32     // This check helps protect against security problems;
  33     // your code file can't be executed directly from the web.
  34     exit(1);
  35 }
  36
  37 /**
  38  * Spam filter class
  39  *
  40  * Local proxy for remote filter
  41  *
  42  * @category  Spam
  43  * @package   StatusNet
  44  * @author    Evan Prodromou <evan@status.net>
  45  * @copyright 2012 StatusNet, Inc.
  46  * @license   http://www.fsf.org/licensing/licenses/agpl-3.0.html AGPL 3.0
  47  * @link      http://status.net/
  48  */
  49
  50 class SpamFilter extends OAuthClient {
  51
  52     const HAM  = 'ham';
  53     const SPAM = 'spam';
  54
  55     public $server;
  56
  57     function __construct($server, $consumerKey, $secret) {
  58         parent::__construct($consumerKey, $secret);
  59         $this->server = $server;
  60     }
  61
  62     protected function toActivity($notice) {
  63         // FIXME: need this to autoload ActivityStreamsMediaLink
  64         $doc = new ActivityStreamJSONDocument();
  65
  66         $activity = $notice->asActivity(null);
  67
  68         return $activity;
  69     }
  70
  71     public function test($notice) {
  72
  73         $activity = $this->toActivity($notice);
  74         return $this->testActivity($activity);
  75     }
  76
  77     public function testActivity($activity) {
  78
  79         $response = $this->postJSON($this->server . "/is-this-spam", $activity->asArray());
  80
  81         $result = json_decode($response->getBody());
  82
  83         return $result;
  84     }
  85
  86     public function train($notice, $category) {
  87
  88         $activity = $this->toActivity($notice);
  89         return $this->trainActivity($activity, $category);
  90
  91     }
  92
  93     public function trainActivity($activity, $category) {
  94
  95         switch ($category) {
  96         case self::HAM:
  97             $endpoint = '/this-is-ham';
  98             break;
  99         case self::SPAM:
 100             $endpoint = '/this-is-spam';
 101             break;
 102         default:
 103             throw new Exception("Unknown category: " + $category);
 104         }
 105
 106         $response = $this->postJSON($this->server . $endpoint, $activity->asArray());
 107
 108         // We don't do much with the results
 109         return true;
 110     }
 111
 112     public function trainOnError($notice, $category) {
 113
 114         $activity = $this->toActivity($notice);
 115
 116         return $this->trainActivityOnError($activity, $category);
 117     }
 118
 119     public function trainActivityOnError($activity, $category) {
 120
 121         $result = $this->testActivity($activity);
 122
 123         if (($category === self::SPAM && $result->isSpam) ||
 124             ($category === self::HAM && !$result->isSpam)) {
 125             return true;
 126         } else {
 127             return $this->trainActivity($activity, $category);
 128         }
 129     }
 130
 131     function postJSON($url, $body)
 132     {
 133         $request = OAuthRequest::from_consumer_and_token($this->consumer,
 134                                                          $this->token,
 135                                                          'POST',
 136                                                          $url);
 137
 138         $request->sign_request($this->sha1_method,
 139                                $this->consumer,
 140                                $this->token);
 141
 142         $hclient = new HTTPClient($url);
 143
 144         $hclient->setConfig(array('connect_timeout' => 120,
 145                                   'timeout' => 120,
 146                                   'follow_redirects' => true,
 147                                   'ssl_verify_peer' => false,
 148                                   'ssl_verify_host' => false));
 149
 150         $hclient->setMethod(HTTP_Request2::METHOD_POST);
 151         $hclient->setBody(json_encode($body));
 152         $hclient->setHeader('Content-Type', 'application/json');
 153         $hclient->setHeader($request->to_header());
 154
 155         // Twitter is strict about accepting invalid "Expect" headers
 156         // No reason not to clear it still here -ESP
 157
 158         $hclient->setHeader('Expect', '');
 159
 160         try {
 161             $response = $hclient->send();
 162             $code = $response->getStatus();
 163             if (!$response->isOK()) {
 164                 throw new OAuthClientException($response->getBody(), $code);
 165             }
 166             return $response;
 167         } catch (Exception $e) {
 168             throw new OAuthClientException($e->getMessage(), $e->getCode());
 169         }
 170     }
 171 }