plugins/ActivitySpam/Spam_score.php

   1 <?php
   2   /**
   3    * StatusNet - the distributed open-source microblogging tool
   4    * Copyright (C) 2011, StatusNet, Inc.
   5    *
   6    * Score of a notice by activity spam service
   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 2011 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     exit(1);
  33 }
  34
  35 /**
  36  * Score of a notice per the activity spam service
  37  *
  38  * @category Spam
  39  * @package  StatusNet
  40  * @author   Evan Prodromou <evan@status.net>
  41  * @license  http://www.fsf.org/licensing/licenses/agpl.html AGPLv3
  42  * @link     http://status.net/
  43  *
  44  * @see      DB_DataObject
  45  */
  46
  47 class Spam_score extends Managed_DataObject
  48 {
  49     const MAX_SCALE = 10000;
  50     public $__table = 'spam_score'; // table name
  51
  52     public $notice_id;   // int
  53     public $score;       // float
  54     public $created;     // datetime
  55
  56     /**
  57      * Get an instance by key
  58      *
  59      * @param string $k Key to use to lookup (usually 'notice_id' for this class)
  60      * @param mixed  $v Value to lookup
  61      *
  62      * @return Spam_score object found, or null for no hits
  63      *
  64      */
  65     function staticGet($k, $v=null)
  66     {
  67         return Managed_DataObject::staticGet('Spam_score', $k, $v);
  68     }
  69
  70     function saveNew($notice, $result) {
  71
  72         $score = new Spam_score();
  73
  74         $score->notice_id      = $notice->id;
  75         $score->score          = $result->probability;
  76         $score->is_spam        = $result->isSpam;
  77         $score->scaled         = Spam_score::scale($score->score);
  78         $score->created        = common_sql_now();
  79         $score->notice_created = $notice->created;
  80
  81         $score->insert();
  82
  83         self::blow('spam_score:notice_ids');
  84
  85         return $score;
  86     }
  87
  88     function save($notice, $result) {
  89
  90         $orig  = null;
  91         $score = Spam_score::staticGet('notice_id', $notice->id);
  92
  93         if (empty($score)) {
  94             $score = new Spam_score();
  95         } else {
  96             $orig = clone($score);
  97         }
  98
  99         $score->notice_id      = $notice->id;
 100         $score->score          = $result->probability;
 101         $score->is_spam        = $result->isSpam;
 102         $score->scaled         = Spam_score::scale($score->score);
 103         $score->created        = common_sql_now();
 104         $score->notice_created = $notice->created;
 105
 106         if (empty($orig)) {
 107             $score->insert();
 108         } else {
 109             $score->update($orig);
 110         }
 111
 112         self::blow('spam_score:notice_ids');
 113
 114         return $score;
 115     }
 116
 117     function delete()
 118     {
 119         self::blow('spam_score:notice_ids');
 120         self::blow('spam_score:notice_ids;last');
 121         parent::delete();
 122     }
 123
 124     /**
 125      * The One True Thingy that must be defined and declared.
 126      */
 127     public static function schemaDef()
 128     {
 129         return array(
 130             'description' => 'score of the notice per activityspam',
 131             'fields' => array(
 132                 'notice_id' => array('type' => 'int',
 133                                      'not null' => true,
 134                                      'description' => 'notice getting scored'),
 135                 'score' => array('type' => 'double',
 136                                  'not null' => true,
 137                                  'description' => 'score for the notice (0.0, 1.0)'),
 138                 'scaled' => array('type' => 'int',
 139                                   'description' => 'scaled score for the notice (0, 10000)'),
 140                 'is_spam' => array('type' => 'tinyint',
 141                                    'description' => 'flag for spamosity'),
 142                 'created' => array('type' => 'datetime',
 143                                    'not null' => true,
 144                                    'description' => 'date this record was created'),
 145                 'notice_created' => array('type' => 'datetime',
 146                                           'description' => 'date the notice was created'),
 147             ),
 148             'primary key' => array('notice_id'),
 149             'foreign keys' => array(
 150                 'spam_score_notice_id_fkey' => array('notice', array('notice_id' => 'id')),
 151             ),
 152             'indexes' => array(
 153                 'spam_score_created_idx' => array('created'),
 154                 'spam_score_scaled_idx' => array('scaled'),
 155             ),
 156         );
 157     }
 158
 159     public static function upgrade()
 160     {
 161         Spam_score::upgradeScaled();
 162         Spam_score::upgradeIsSpam();
 163         Spam_score::upgradeNoticeCreated();
 164     }
 165
 166     protected static function upgradeScaled()
 167     {
 168         $score = new Spam_score();
 169         $score->whereAdd('scaled IS NULL');
 170
 171         if ($score->find()) {
 172             while ($score->fetch()) {
 173                 $orig = clone($score);
 174                 $score->scaled = Spam_score::scale($score->score);
 175                 $score->update($orig);
 176             }
 177         }
 178     }
 179
 180     protected static function upgradeIsSpam()
 181     {
 182         $score = new Spam_score();
 183         $score->whereAdd('is_spam IS NULL');
 184
 185         if ($score->find()) {
 186             while ($score->fetch()) {
 187                 $orig = clone($score);
 188                 $score->is_spam = ($score->score >= 0.90) ? 1 : 0;
 189                 $score->update($orig);
 190             }
 191         }
 192     }
 193
 194     protected static function upgradeNoticeCreated()
 195     {
 196         $score = new Spam_score();
 197         $score->whereAdd('notice_created IS NULL');
 198
 199         if ($score->find()) {
 200             while ($score->fetch()) {
 201                 $notice = Notice::staticGet('id', $score->notice_id);
 202                 if (!empty($notice)) {
 203                     $orig = clone($score);
 204                     $score->notice_created = $notice->created;
 205                     $score->update($orig);
 206                 }
 207             }
 208         }
 209     }
 210
 211     public static function scale($score)
 212     {
 213         $raw = round($score * Spam_score::MAX_SCALE);
 214         return max(0, min(Spam_score::MAX_SCALE, $raw));
 215     }
 216 }