[Sommaire] [Imprimer]

   1  <?php
   2  
   3      lt_include( PLOG_CLASS_PATH."class/database/dbobject.class.php" );
   4  
   5      define("TOKEN_DEFAULT_PROBABILITY", 0.4);
   6      
   7      define("TOKEN_SEPARATOR_MARK", "#");
   8      define("TOKEN_TOPIC_MARK", "Topic" . TOKEN_SEPARATOR_MARK);
   9      define("TOKEN_URL_MARK", "Url" . TOKEN_SEPARATOR_MARK);
  10      define("TOKEN_USER_NAME_MARK", "UserName" . TOKEN_SEPARATOR_MARK);
  11      define("TOKEN_USER_EMAIL_MARK", "UserEmail" . TOKEN_SEPARATOR_MARK);
  12      define("TOKEN_USER_URL_MARK", "UserUrl" . TOKEN_SEPARATOR_MARK);
  13      
  14      /**
  15       * \ingroup DAO
  16       *
  17       * Represents a record form the plog_filtered_content table
  18       *
  19       * The key of this class is the regexp that will be used to match
  20       * words against it.
  21       */
  22      class BayesianToken extends DbObject 
  23      {
  24          
  25          var $_id;
  26          var $_blogId;
  27          var $_token;
  28          var $_spamOccurrences;
  29          var $_nonSpamOccurrences;
  30          var $_prob;
  31          
  32          /**
  33          * -- Add function info here --
  34          */
  35      	function BayesianToken($blogId, $token, $spamOccurrences, $nonSpamOccurrences, $prob, $id = -1)
  36          {
  37              $this->DbObject();
  38  
  39              $this->_id                  = $id;
  40              $this->_blogId              = $blogId;
  41              $this->_token               = $token;
  42              $this->_spamOccurrences     = $spamOccurrences;
  43              $this->_nonSpamOccurrences  = $nonSpamOccurrences;
  44              $this->_prob                = $prob;
  45          }
  46          
  47          /**
  48          * -- Add function info here --
  49          */
  50          function getId()
  51          {
  52              return $this->_id;
  53          }
  54          
  55          /**
  56          * -- Add function info here --
  57          */
  58          function getBlogId()
  59          {
  60              return $this->_blogId;
  61          }
  62          
  63          /**
  64          * -- Add function info here --
  65          */
  66          function getToken()
  67          {
  68              return $this->_token;
  69          }
  70          
  71          /**
  72          * -- Add function info here --
  73          */
  74          function getSpamOccurrences()
  75          {
  76              return $this->_spamOccurrences;
  77          }
  78          
  79          /**
  80          * -- Add function info here --
  81          */
  82          function getNonSpamOccurrences()
  83          {
  84              return $this->_nonSpamOccurrences;
  85          }
  86          
  87          /**
  88          * -- Add function info here --
  89          */
  90          function getProb()
  91          {
  92              return $this->_prob;
  93          }
  94          
  95          /**
  96          * -- Add function info here --
  97          */
  98          function setProb($prob)
  99          {
 100              $this->_prob = $prob;
 101          }
 102          
 103          /**
 104          * -- Add function info here --
 105          */
 106          function isSignificant()
 107          {            
 108              return ((2 * $this->getNonSpamOccurrences()) + $this->getSpamOccurrences()) > 5;
 109          }
 110          
 111          /**
 112          * -- Add function info here --
 113          */
 114          function isValid($token = false)
 115          {
 116              $config    =& Config::getConfig();
 117              
 118              if ($token === false)
 119              {
 120                  $token = $this->getToken();
 121              }
 122              
 123              $valid = !ereg("^[0-9-]+$", $token);
 124  
 125              if ($config->getValue("bayesian_filter_min_length_token") > 0)
 126              {                                                
 127                  $valid = $valid && strlen($token) >= $config->getValue("bayesian_filter_min_length_token");
 128              }
 129              
 130              if ($config->getValue("bayesian_filter_max_length_token") > 0)
 131              {
 132                  $valid = $valid && strlen($token) <= $config->getValue("bayesian_filter_max_length_token");
 133              }
 134              
 135              return $valid;
 136          }
 137          
 138          /**
 139          * -- Add function info here --
 140          */
 141          function degenerate($token = false)
 142          {
 143              if ($token === false)
 144              {
 145                  $token = $this->getToken();
 146              }
 147     
 148              $degenerations = array();
 149              $curToken = $token;
 150              $prefix = true;
 151              
 152              while ($prefix)
 153              {
 154                  if (ereg("^(" . TOKEN_TOPIC_MARK . "|" . 
 155                                  TOKEN_URL_MARK . "|" . 
 156                                  TOKEN_USER_NAME_MARK . "|" .
 157                                  TOKEN_USER_EMAIL_MARK . "|" .
 158                                  TOKEN_USER_URL_MARK . ")(.*)$", $curToken, $regs))
 159                  {
 160                      $curToken = $regs[2];
 161                      $prefix = $regs[1];
 162                  }
 163                  else
 164                  {
 165                      $prefix = "";
 166                  }
 167  
 168                  $degenerations = array_merge($degenerations, BayesianToken::getBasicsDegeneration($curToken, $prefix));
 169  
 170                  if (ereg("([^!]+!)!+$", $curToken, $regs))
 171                  {
 172                      $degenerations = array_merge($degenerations, BayesianToken::getBasicsDegeneration($regs[1], $prefix));
 173                  }
 174                  
 175                  if (ereg("([^!]+)!+$", $curToken, $regs))
 176                  {   
 177                      $degenerations = array_merge($degenerations, BayesianToken::getBasicsDegeneration($regs[1], $prefix));
 178                  }   
 179                  
 180                  if ($prefix == "" && $curToken == $token)
 181                  {
 182                      foreach ($degenerations as $degeneration)
 183                      {
 184                          array_push($degenerations, TOKEN_TOPIC_MARK . $degeneration);
 185                          array_push($degenerations, TOKEN_URL_MARK . $degeneration);
 186                          array_push($degenerations, TOKEN_USER_NAME_MARK . $degeneration);
 187                          array_push($degenerations, TOKEN_USER_EMAIL_MARK . $degeneration);
 188                          array_push($degenerations, TOKEN_USER_URL_MARK . $degeneration);                        
 189                      }
 190                  }
 191              }
 192  
 193              return array_unique($degenerations);
 194          }
 195          
 196          /**
 197          * -- Add function info here --
 198          */
 199          function getBasicsDegeneration($token = false, $prefix = "")
 200          {
 201              if ($token === false)
 202              {
 203                  $token = $this->getToken();
 204              }
 205              
 206              $degenerations = array();
 207              
 208              $lower = strtolower($token);                 
 209              array_push($degenerations, $prefix . $token);
 210              array_push($degenerations, $prefix . strtoupper($token));
 211              array_push($degenerations, $prefix . $lower);
 212              array_push($degenerations, $prefix . ucFirst($lower));
 213  
 214              return $degenerations;
 215          }
 216      }
 217  ?>