CMailFilter::GetSpamRating

static function GetSpamRating($message)
{
	global $DB;

	$arWords = CMailFilter::getWords($message, 1000);

	if (empty($arWords))
		return 0;

	// for every word find Si
	$arWords = array_map("md5", $arWords);

	global $BX_MAIL_SPAM_CNT;
	if(!is_set($BX_MAIL_SPAM_CNT, "G"))
	{
		$strSql = "SELECT MAX(GOOD_CNT) as G, MAX(BAD_CNT) as B FROM b_mail_spam_weight";
		if($res = $DB->Query($strSql))
			$BX_MAIL_SPAM_CNT = $res->Fetch();

		if(intval($BX_MAIL_SPAM_CNT["G"])<=0)
			$BX_MAIL_SPAM_CNT["G"] = 1;

		if(intval($BX_MAIL_SPAM_CNT["B"])<=0)
			$BX_MAIL_SPAM_CNT["B"] = 1;
	}

	$CNT_WORDS = COption::GetOptionInt("mail", "spam_word_count", B_MAIL_WORD_CNT);
	$MIN_COUNT =  COption::GetOptionInt("mail", "spam_min_count", B_MAIL_MIN_CNT);
	// select $CNT_WORDS words with max |Si - 0.5|
	// if the word placed less then xxx (5) times, then ignore
	$strSql =
		"SELECT SW.*, ".
		"	(BAD_CNT/".$BX_MAIL_SPAM_CNT["B"].".0) / (2*GOOD_CNT/".$BX_MAIL_SPAM_CNT["G"].".0 + BAD_CNT/".$BX_MAIL_SPAM_CNT["B"].".0) as RATING, ".
		"	ABS((BAD_CNT/".$BX_MAIL_SPAM_CNT["B"].".0) / (2*GOOD_CNT/".$BX_MAIL_SPAM_CNT["G"].".0 + BAD_CNT/".$BX_MAIL_SPAM_CNT["B"].".0) - 0.5) as MOD_RATING ".
		"FROM b_mail_spam_weight SW ".
		"WHERE WORD_ID IN ('".implode("', '", $arWords)."') ".
		"	AND ABS((BAD_CNT/".$BX_MAIL_SPAM_CNT["B"].".0) / (2*GOOD_CNT/".$BX_MAIL_SPAM_CNT["G"].".0 + BAD_CNT/".$BX_MAIL_SPAM_CNT["B"].".0) - 0.5) > 0.1 ".
		"	AND TOTAL_CNT>".$MIN_COUNT." ".
		"ORDER BY MOD_RATING DESC ".
		($DB->type == "MYSQL"?"LIMIT ".$CNT_WORDS : "");

	//echo htmlspecialcharsbx($strSql)."
";

	$a = 1;
	$b = 1;
	$dbr = $DB->Query($strSql, false, "File: ".__FILE__."
Line: ".__LINE__);
	$arr = true;
	$words = "";

	for($i=0; $i<$CNT_WORDS; $i++)
	{
		if($arr && $arr = $dbr->Fetch())
		{
			//echo "".htmlspecialcharsbx($arr["WORD_REAL"])."=".$arr["RATING"]."
 ";
			$words .= $arr["WORD_REAL"]." ".Round($arr["RATING"]*100, 4)." ".$arr["BAD_CNT"]." ".$arr["GOOD_CNT"]."n";
			$a = $a * ($arr["RATING"]==0?0.00001:$arr["RATING"]);
			$b = $b * (1 - ($arr["RATING"]==1?0.9999:$arr["RATING"]));
		}
		else
		{
			//if there is no word then weight Si = 0.4
			$a = $a * 0.4;
			$b = $b * (1 - 0.4);
		}
	}
	// calculate Bayes for the whole message
	$rating = $a/($a+$b) * 100;

	return Array("RATING"=>$rating, "WORDS"=>$words);
}