You are viewing a plain text version of this content. The canonical link for it is here.
Posted to commits@spamassassin.apache.org by qu...@apache.org on 2004/03/17 09:26:20 UTC
svn commit: rev 9564 - in incubator/spamassassin/trunk: lib/Mail/SpamAssassin rules
Author: quinlan
Date: Wed Mar 17 00:26:20 2004
New Revision: 9564
Modified:
incubator/spamassassin/trunk/lib/Mail/SpamAssassin/HTML.pm
incubator/spamassassin/trunk/rules/70_testing.cf
Log:
fix comment ratio code
test new thresholds for HTML_COMMENT_RATIO test
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 Wed Mar 17 00:26:20 2004
@@ -1052,11 +1052,12 @@
push @{$self->{html_text}}, $text;
}
+# note: $text includes <!-- and -->
sub html_comment {
my ($self, $text) = @_;
push @{ $self->{html}{comment} }, $text;
- $self->{html}{total_comment_length} += length($text) + 7; # "<!--" + "-->"
+ $self->{html}{total_comment_length} += length($text);
if ($self->{html_last_tag} eq "div" &&
$text =~ /Converted from text\/plain format/)
Modified: incubator/spamassassin/trunk/rules/70_testing.cf
==============================================================================
--- incubator/spamassassin/trunk/rules/70_testing.cf (original)
+++ incubator/spamassassin/trunk/rules/70_testing.cf Wed Mar 17 00:26:20 2004
@@ -409,9 +409,6 @@
body T_PIPE_OBFU /[a-z]\|[a-z]/i
describe T_PIPE_OBFU Words dis|guised with a pipe sym|bol
-body T_HTML_COMMENT_TAG eval:html_text('comment', '=~ /<.*?>/')
-describe T_HTML_COMMENT_TAG HTML comment probably contains HTML tag
-
# Should catch Atriks, for one
uri T_LONG_HOSTNAME_25 /^https?:\/\/[a-z0-9]{25,29}\./i
uri T_LONG_HOSTNAME_30 /^https?:\/\/[a-z0-9]{30,40}\./i
@@ -725,3 +722,25 @@
body T_HTML_IMG_ENDS_SLASH eval:html_text('img_src', '=~ m{/$}')
body T_HTML_WEB_BUGS eval:html_text('img_src', '=~ /\.(?:pl|cgi|php|asp|jsp|cfm)\b/i')
describe T_HTML_WEB_BUGS Image tag intended to identify you
+
+# revisit comment ratio
+body __HTML_COMMENT_RATIO_75 eval:html_range('total_comment_ratio','0.75','1.00')
+meta HTML_COMMENT_RATIO_75 __HTML_COMMENT_RATIO && MIME_HTML_ONLY
+describe HTML_COMMENT_RATIO_75 HTML comments are large percentage of message
+
+body __HTML_COMMENT_RATIO_80 eval:html_range('total_comment_ratio','0.80','1.00')
+meta HTML_COMMENT_RATIO_80 __HTML_COMMENT_RATIO && MIME_HTML_ONLY
+describe HTML_COMMENT_RATIO_80 HTML comments are large percentage of message
+
+body __HTML_COMMENT_RATIO_85 eval:html_range('total_comment_ratio','0.85','1.00')
+meta HTML_COMMENT_RATIO_85 __HTML_COMMENT_RATIO && MIME_HTML_ONLY
+describe HTML_COMMENT_RATIO_85 HTML comments are large percentage of message
+
+body __HTML_COMMENT_RATIO_90 eval:html_range('total_comment_ratio','0.90','1.00')
+meta HTML_COMMENT_RATIO_90 __HTML_COMMENT_RATIO && MIME_HTML_ONLY
+describe HTML_COMMENT_RATIO_90 HTML comments are large percentage of message
+
+body __HTML_COMMENT_RATIO_95 eval:html_range('total_comment_ratio','0.95','1.00')
+meta HTML_COMMENT_RATIO_95 __HTML_COMMENT_RATIO && MIME_HTML_ONLY
+describe HTML_COMMENT_RATIO_95 HTML comments are large percentage of message
+