Merge branch 'master' of github.com:courtenay/splam

Conflicts:
	lib/splam/rule.rb
	lib/splam/rules/bad_words.rb
	lib/splam/rules/good_words.rb
	lib/splam/rules/line_length.rb

lib/splam/rule.rb

@@ -32,6 +32,7 @@ def run(*args)
 
   def initialize(suite, record, weight = 1.0, request = nil)
     @suite, @weight, @score, @reasons, @body, @request = suite, weight, 0, [], record.send(suite.body), request
+    @user = record.user # todo: customize user field
   end
 
   def name

lib/splam/rules/bad_words.rb

@@ -66,6 +66,7 @@ def run
           @body.scan(/<a(.*?)>/).each do |match|
             add_score self.class.bad_word_score * 10 * match[0].scan(word).size, "nasty word inside a URL: #{word}"
           end
+
         end
         if counter > (wordlist.size / 2)
           add_score 1000, "Lots of bad words from one genre (#{key}): #{counter}"

lib/splam/rules/chinese.rb

@@ -2,7 +2,7 @@ class Splam::Rules::Chinese < Splam::Rule
   class << self
     attr_accessor :base_score
   end
-  self.base_score = 3
+  self.base_score = 5
 
   def run
     banned_words =[ # various chinese characters

lib/splam/rules/good_words.rb

@@ -7,8 +7,8 @@ def run
     good_words |= %w( redirect login diff dreamhost setup subversion git  wildcard domain subdomain ssh database )
     good_words |= %w( project billing tags description comment milestone saving happening feature mac implement report)
     good_words |= %w( rss notification subscribe calendar chart note task gantt search service ownership application communicate )
-    good_words |= %w( pattern template web integer status xml activereload html state page)
     good_words |= %w( interaction API tickets hosted domain skitch )
+    good_words |= %w( pattern template web integer status xml activereload html state page rack diff )
     good_words << "project management"
     good_words << "/usr/local/lib" << "gems"
 

lib/splam/rules/line_length.rb

@@ -0,0 +1,26 @@
+class Splam::Rules::LineLength < Splam::Rule
+
+  def name
+    "Line length"
+  end
+
+  # Penalize long line lengths.
+  def run
+    lines = @body.split("\n")
+    lines.each do |line|
+      next if line =~ /\A\s{4,}/ # ignore code blocks
+
+      # multiplier = (lines.size == 1) ? 10 : 1 # one line? fail.
+      multiplier = 1
+
+      # 1 point for each 40 chars in a line.
+      hits = (line.size / 40) * multiplier
+      add_score hits, "lines over 40 chars"
+
+      # 2 more points if line is longer than 80
+      hits = (line.size / 80) * 2 * multiplier
+      add_score hits, "lines over 80 chars"
+
+    end
+  end
+end

lib/splam/rules/user.rb

@@ -0,0 +1,15 @@
+class Splam::Rules::User < Splam::Rule
+
+  def run
+    bad_words = ["qq.com", "yahoo.cn", "126.com"]
+    bad_words |= %w( mortgage )
+
+    bad_words.each do |word|
+      add_score 50, "User's email address has suspicious parts: #{word}" if @user.email.include?(word)
+    end
+
+    add_score "20", "User has lots and lots of dots" if @user.email.split("@")[0].scan(/\./).size > 5
+
+    add_score 5, "User is untrusted" if !@user.trusted?
+  end
+end

lib/splam/rules/word_length.rb

@@ -16,17 +16,16 @@ def median(arr)
 
   def run
     words = []
-    words = @body.split(/\s/).map do |word|
-      word.size
-    end
-    words.delete_if { |w| w =~ /^http\:\/\//}
+    words = @body.split(/\s/)
+    words.delete_if { |w| w =~ /^https?\:\/\// }
+    words.collect! { |word| word.size }
 
     # Only count word lengths over 10
     if words.size > 5
-      add_score 20, "Average word length over 5"  if average(words) > 5
-      add_score 50, "Average word length over 10" if average(words) > 10
-      add_score 10, "Median word length over 5"   if median(words) > 5
-      add_score 50, "Median word length over 10"  if median(words) > 10
+      add_score 5, "Average word length over 5"  if average(words) > 5
+      add_score 10, "Average word length over 10" if average(words) > 10
+      add_score 5, "Median word length over 5"   if median(words) > 5
+      add_score 10, "Median word length over 10"  if median(words) > 10
     end
   end
 end

test/splam_test.rb

@@ -7,13 +7,26 @@ def run
     end
   end
 
+  class User
+    attr_accessor :trusted
+    def trusted?
+      trusted
+    end
+    def email
+      "test@test.com"
+    end
+  end
+
   class Foo
     include ::Splam
     splammable :body
     attr_accessor :body
     def body
       @body || "This is body\320\224 \320\199"
     end
+    def user
+      User.new
+    end
   end
 
   class FooCond
@@ -27,7 +40,9 @@ class PickyFoo
     splammable :body do |s|
       s.rules = [:fixed_rule, FixedRule]
     end
-
+    def user
+      User.new
+    end
     def body
       'lol wut'
     end
@@ -38,6 +53,9 @@ class HeavyFoo
     splammable :body do |s|
       s.rules = {:fixed_rule => 3}
     end
+    def user
+      User.new
+    end
 
     def body
       'lol wut'
@@ -47,7 +65,7 @@ def body
   def test_runs_plugins
     f = Foo.new
     assert ! f.splam?
-    assert_equal 35, f.splam_score
+    assert_equal 40, f.splam_score
   end
 
   def test_runs_plugins_with_specified_rules