[Spamscan-cvs] spamscan/bin spamscan,1.4,1.5

[<de...@us...>]

Update of /cvsroot/spamscan/spamscan/bin
In directory sc8-pr-cvs1:/tmp/cvs-serv32083

Modified Files:
	spamscan 
Log Message:
Added additional code to check in db for words on incoming spam.
This needs to be configurable in the users' spamscan.conf file.



Index: spamscan
===================================================================
RCS file: /cvsroot/spamscan/spamscan/bin/spamscan,v
retrieving revision 1.4
retrieving revision 1.5
diff -C2 -d -r1.4 -r1.5
*** spamscan	22 Jul 2003 20:52:25 -0000	1.4
--- spamscan	23 Jul 2003 02:23:05 -0000	1.5
***************
*** 148,151 ****
--- 148,152 ----
  &to_filter;
  &body_filter;
+ &db_filter;
  
  if ($spampassword eq $passcode) {
***************
*** 295,299 ****
  sub spam_algorithm {
          # super lame right now
!         $spam_weight = $from_value + $to_value + $body_value;
          if ($spam_weight >= $spam_threshold) {
                  $itsspam = 1;
--- 296,300 ----
  sub spam_algorithm {
          # super lame right now
!         $spam_weight = $from_value + $to_value + $body_value + $dbcount;
          if ($spam_weight >= $spam_threshold) {
                  $itsspam = 1;
***************
*** 317,321 ****
                  $itsspam = 0;
          }
!         $spamreason_msg = "X-Spam-Scanned: Weight Totals:  TOTAL: $spam_weight - From: $from_value\, To: $to_value\, Body: $body_value\n";
  	foreach $reason (@spamreasons) {
  		$spamreason_msg = $spamreason_msg . "X-Spam-Scanned: $reason\n";
--- 318,322 ----
                  $itsspam = 0;
          }
!         $spamreason_msg = "X-Spam-Scanned: Weight Totals:  TOTAL: $spam_weight - From: $from_value\, To: $to_value\, Body: $body_value\, DB: $dbcount\n";
  	foreach $reason (@spamreasons) {
  		$spamreason_msg = $spamreason_msg . "X-Spam-Scanned: $reason\n";
***************
*** 500,501 ****
--- 501,536 ----
  	@completemailmsg = @newcompletemailmsg;
  }
+ 
+ sub db_filter {
+ 	$dbfile = $filterdir . "spamscan";
+         dbmopen(%phrasedb, "$dbfile", 0644) or die "dbmopen: $! ($dbfile)";
+ 	# first, check the from address.  If it exists in the db, then it's a non-spam address. 
+ 	if ($phrasedb{$fromemail}) {
+ 		$dbcount = $dbcount - 20;
+ 	}
+ 	# now look through phrases, and total up count.
+ 	@words = split/\s+/,$learntext;
+ 
+         $maxdepth = 4;
+         $maxsize = ($#words + 1);
+ 
+         # this loop does all the 1,2, or 3 word combos etc
+         for ($i=0; $i < $maxdepth; $i++) {
+                 # this is the "starting element" loop
+                 for ($j=0; $j < $maxsize; $j++) {
+                         $end = $j + $i;
+                         # this would be the printing loop
+                         for ($k=$j; $k <= $end; $k++) {
+                                 $wordling = "$wordling " . "$words[$k]";
+                         }
+                         $wordling =~ s/\s+$//;
+                         $wordcksum = unpack ("%32C*", $wordling) % 32767;
+                         # $phrasedb{$wordcksum}--;
+                         $dbcount = $dbcount + $phrasedb{$wordling};
+                         $wordling = "";
+                 }
+         }
+ 
+ }
+ 
+