test/fixtures/comment/spam/rapid.txt test/fixtures/comment/spam/tk.txt @@ -9,12 +9,12 @@ class Splam::Rules::BadWords < Splam::Rule def run bad_words = %w( sex sexy porn gay erotica viagra erotismo porno porn lesbian amateur tit) bad_words |= %w( gratis erotismo porno torrent bittorrent adulto ) - bad_words |= %w( cialis viagra payday loan ) - bad_words |= %w( webcam free-web-host) + bad_words |= %w( cialis viagra payday loan jihad ) + bad_words |= %w( webcam free-web-host rapidshare muslim) bad_words << /pel?cula/ << /pornogr?fica/ << "portal porno" # srsly, spamming in spanish? - suspicious_words = %w( free buy galleries dating gallery hard hardcore video homemade celebrity ) << "credit card" - suspicious_words |= %w( adult pharmacy overnight shipping free hot movie nylon arab ?????? xxx) + suspicious_words = %w( free buy galleries dating gallery hard hardcore video homemade celebrity ) << "credit card" << "my friend" << "friend sent me" + suspicious_words |= %w( adult pharmacy overnight shipping free hot movie nylon arab ?????? xxx) << "sent me a link" suspicious_words << "forums/member.php?u=" << "chat room" << "free chat" << "yahoo chat" << "page.php" bad_words.each do |word| results = @body.downcase.scan(word) @@ -24,6 +24,9 @@ class Splam::Rules::BadWords < Splam::Rule @body.scan(/<a[^>]+>(.*?)<\/a>/).each do |match| add_score self.class.bad_word_score * 4 * match[0].scan(word).size, "nasty word inside a link: #{word}" end + @body.scan(/\nhttp:\/\/(.*?#{word})/).each do |match| + add_score self.class.bad_word_score ** 4 * match[0].scan(word).size, "nasty word inside a straight-up link: #{word}" + end @body.scan(/<a(.*?)>/).each do |match| add_score self.class.bad_word_score * 4 * match[0].scan(word).size, "nasty word inside a URL: #{word}" end lib/splam/rules/bad_words.rb @@ -21,6 +21,7 @@ class Splam::Rules::Href < Splam::Rule 'cn' => 20, # Chinese? spammer. 'us' => 8, # .us ? possibly spam 'it' => 5, + 'tk' => 20, 'pl' => 8, 'info' => 20, 'biz' => 40 # no-one uses these for reals @@ -32,8 +33,8 @@ class Splam::Rules::Href < Splam::Rule tokens = @body.split(" ") if tokens[-1] =~ /^http:\/\// - add_score 50, "Text ends in a link" - add_score 10, "Text ends in a link and only has one link" if link_count == 1 + add_score 10, "Text ends in a http token" + add_score 50, "Text ends in a http token and only has one token" if link_count == 1 end @body.scan(/http:\/\/(.*?)[\/\]?]/) do |match| lib/splam/rules/href.rb @@ -81,7 +81,7 @@ class SplamTest < Test::Unit::TestCase end spam = comment.splam? score = comment.splam_score - #$stderr.puts "#{f} score: #{score}" + #$stderr.puts "#{f} score: #{score}\n#{comment.splam_reasons.inspect}" #$stderr.puts "=====================" assert spam, "Comment #{f} was not spam, score was #{score} but threshold was #{Foo.splam_suite.threshold}\nReasons were #{comment.splam_reasons.inspect}" end test/splam_test.rb bbe48ee06e01fbe926fe34e76ad2019a4be9d79c courtenay court3nay@gmail.com http://github.com/courtenay/splam/commit/412d8e178c1daa2d09247e2c5c38b49ec0e0280b 412d8e178c1daa2d09247e2c5c38b49ec0e0280b 2009-02-06T10:32:33-08:00 2009-02-06T10:32:33-08:00 Add some more bad words and spam that slipped through the cracks. 46316e9241a0d5ea6b73bd95e3d63f4ed3242963 courtenay court3nay@gmail.com