svn commit: rev 6463 - in incubator/spamassassin/trunk: lib/Mail/SpamAssassin rules

[qu...@apache.org]

Author: quinlan
Date: Tue Feb  3 16:29:53 2004
New Revision: 6463

Modified:
   incubator/spamassassin/trunk/lib/Mail/SpamAssassin/Dns.pm
   incubator/spamassassin/trunk/lib/Mail/SpamAssassin/HTML.pm
   incubator/spamassassin/trunk/rules/70_cvs_rules_under_test.cf
Log:
add HTML parser-based version of T_HTML_CONVERTED
add xbl+sbl support
a few other new test rules


Modified: incubator/spamassassin/trunk/lib/Mail/SpamAssassin/Dns.pm
==============================================================================
--- incubator/spamassassin/trunk/lib/Mail/SpamAssassin/Dns.pm	(original)
+++ incubator/spamassassin/trunk/lib/Mail/SpamAssassin/Dns.pm	Tue Feb  3 16:29:53 2004
@@ -263,8 +263,12 @@
       }
     }
     # regular expression
-    elsif ($rdatastr =~ /\Q$subtest\E/) {
-      $self->dnsbl_hit($rule, $question, $answer);
+    else {
+      my $test = qr/$subtest/;
+      if ($rdatastr =~ /$test/) {
+	print STDERR "here for $rule $question $answer\n";
+	$self->dnsbl_hit($rule, $question, $answer);
+      }
     }
   }
 }

Modified: incubator/spamassassin/trunk/lib/Mail/SpamAssassin/HTML.pm
==============================================================================
--- incubator/spamassassin/trunk/lib/Mail/SpamAssassin/HTML.pm	(original)
+++ incubator/spamassassin/trunk/lib/Mail/SpamAssassin/HTML.pm	Tue Feb  3 16:29:53 2004
@@ -719,6 +719,11 @@
   $self->{html}{comment_text} .= "$text\n";
   $self->{html}{total_comment_length} += length($text) + 7; # "<!--" + "-->"
 
+  if ($self->{html_last_tag} eq "div" &&
+      $text =~ /Converted from text\/plain format/)
+  {
+    $self->{html}{t_html_converted} = 1;
+  }
   if (exists $self->{html}{"inside_script"} && $self->{html}{"inside_script"} > 0)
   {
     if ($text =~ /\b(?:$events)\b/io)

Modified: incubator/spamassassin/trunk/rules/70_cvs_rules_under_test.cf
==============================================================================
--- incubator/spamassassin/trunk/rules/70_cvs_rules_under_test.cf	(original)
+++ incubator/spamassassin/trunk/rules/70_cvs_rules_under_test.cf	Tue Feb  3 16:29:53 2004
@@ -275,7 +275,8 @@
 header T_FAKE_ENVFROM_ALTAVISTA	EnvelopeFrom =~ /\@altavista\.com$/i
 
 header T_FAKE_OUTBLAZE_RCVD	Received =~ /\.mr\.outblaze\.com/
-header T_FAKE_OUTBLAZE_MID	Message-Id =~ /\.mr\.outblaze\.com/
+header T_FAKE_OUTBLAZE_MID_1	Message-Id =~ /\.mr\.outblaze\.com/
+header T_FAKE_OUTBLAZE_MID_2	MESSAGEID =~ /\.mr\.outblaze\.com/
 
 header T_ALL_TRUSTED		eval:check_all_trusted()
 describe T_ALL_TRUSTED		Did not pass through any untrusted hosts
@@ -295,11 +296,21 @@
 rawbody T_BAYESBUSTER_LINE_15	/^([a-z]{3,} ){15,}<[Bb][Rr]>$/
 rawbody T_BAYESBUSTER_LINE_15I	/^([a-z]{3,} ){15,}<br>$/i
 
+# XBL+SBL
+header __RCVD_IN_SBL_XBL	eval:check_rbl_txt('sblxbl', 'sbl-xbl.spamhaus.org.')
+describe __RCVD_IN_SBL_XBL	Received via a relay in Spamhaus SBL+XBL
+tflags __RCVD_IN_SBL_XBL	net
+
+# SBL replacement
+header T_RCVD_IN_SBL	eval:check_rbl_sub('sblxbl', '(?i)/sbl')
+describe T_RCVD_IN_SBL	Received via a relay in Spamhaus SBL
+tflags T_RCVD_IN_SBL	net
+
 # XBL is the Exploits Block List: http://www.spamhaus.org/xbl/
 # contains data from the CBL, probably supercedes that.
-header T_RCVD_IN_XBL      eval:check_rbl_txt('xbl', 'xbl.spamhaus.org.')
-describe T_RCVD_IN_XBL    Received via a relay in Spamhaus XBL
-tflags T_RCVD_IN_XBL      net
+header T_RCVD_IN_XBL	eval:check_rbl_sub('sblxbl', '(?i)/xbl')
+describe T_RCVD_IN_XBL	Received via a relay in Spamhaus XBL
+tflags T_RCVD_IN_XBL	net
 
 # reported by Kurtis Rader, bug 2890
 uri T_MSN_REDIR		/^http:\/\/shopping\.msn\.com\/trackurl\.aspx\?/i
@@ -454,12 +465,17 @@
 header T_MIME_BOUND_DIGITS15	Content-Type =~ /boundary=\"\d{15,}\"/
 describe T_MIME_BOUND_DIGITS15	MIME boundary contains all digits
 
-header T_MSGID_SPAM_2		Message-Id =~ /<[A-Z]{7}-000[0-9]{10}\@[a-z]*>/
-describe T_MSGID_SPAM_2		Message-ID has known spammer pattern
+header T_MSGID_SPAM_2_A		Message-Id =~ /<[A-Z]{7}-000[0-9]{10}\@[a-z]*>/
+describe T_MSGID_SPAM_2_A	Message-ID has known spammer pattern
+header T_MSGID_SPAM_2_B		MESSAGEID =~ /<[A-Z]{7}-000[0-9]{10}\@[a-z]*>/
+describe T_MSGID_SPAM_2_B	Message-ID has known spammer pattern
 
 # Comment is a spam sign when following <DIV>
 rawbody T_HTML_CONVERTED	m{<DIV><!-- Converted from text/plain format -->}
 describe T_HTML_CONVERTED	HTML conversion tool used by spam
+
+body T_HTML_CONVERTED_2		eval:html_test('t_html_converted')
+describe T_HTML_CONVERTED_2	HTML conversion tool used by spam
 
 # another Message-Id format (pick one as MSGID_SPAM_3)
 header T_MSGID_SPAM_3_5		Message-Id =~ /<[a-z]{5,}\@(\S+\.)+\S+>/