Added the SPAMINATOR!

whatupdave · whatupdave · commit 2d2a3bc58b65 · 2018-01-11T20:09:58.000-08:00
diff --git a/app/controllers/protips_controller.rb b/app/controllers/protips_controller.rb
@@ -195,34 +195,13 @@ def etag_key_for_protip
   end
 
   def spam?
-    is_spam = false
-    if smyte_spam?
-      is_spam = true
-      logger.info "[SMYTE-SPAM BLOCK] \"#{@protip.title}\""
+    flags = Spaminator.new.protip_flags(@protip)
+    if flags.any?
+      logger.info "[SPAM BLOCK] \"#{@protip.title}\"  #{flags.inspect}"
+      true
     else
-      logger.info "[SMYTE-SPAM ALLOW] \"#{@protip.title}\""
+      logger.info "[SPAM ALLOW] \"#{@protip.title}\""
+      false
     end
-
-    if @protip.looks_spammy?
-      is_spam = true
-      logger.info "[CW-SPAM BLOCK] \"#{@protip.title}\""
-    else
-      logger.info "[CW-SPAM ALLOW] \"#{@protip.title}\""
-    end
-
-    is_spam
-  end
-
-  def smyte_spam?
-    return false if ENV['SMYTE_URL'].nil?
-    data = {
-      actor: serialize(current_user, CurrentUserSerializer),
-      protip: serialize(@protip).except("spam_detected_at", "bad_content")
-    }
-    Smyte.new.spam?(
-      'post_protip',
-      data,
-      request
-    )
   end
 end
diff --git a/app/lib/spaminator.rb b/app/lib/spaminator.rb
@@ -0,0 +1,66 @@
+URLS = /(^$)|(^(http|https):\/\/[a-z0-9]+([\-\.]{1}[a-z0-9]+)*\.[a-z]{2,5}(([0-9]{1,5})?\/.*)?$)/ix
+
+class Spaminator
+  def bad_links?(text, urls)
+    text.scan(/shurll.com|shorl.com/i).size > 1
+  end
+
+  def recognized_format?(text)
+    text.match(/^\[\!\[Foo\]/)
+  end
+
+  def customer_support?(text)
+    text.scan(/customer|support|phonenumber|phonesupport/i).size > 10
+  end
+
+  def download_links?(text, urls, title)
+    title.match(/serial key|free download/i) ||
+      text.scan(/download|crack|serial|torrent/i).size > 10
+  end
+
+  def many_spaces?(text, urls, title)
+    title.scan(/ /).size > 2
+  end
+
+  def mostly_url?(text, urls)
+    urls.join.size / text.size.to_f > 0.5
+  end
+
+  def weird_characters?(text)
+    text.scan(/[\.]/).size / text.size.to_f > 0.10
+  end
+
+  def protip_flags(protip)
+    flags = []
+    text = [protip.title, protip.body, protip.tags].flatten.join("\n")
+    urls = URI.extract(text).compact
+
+    flags << 'bad_user' if protip.user.bad_user
+    flags << 'bad_links' if bad_links?(text, urls)
+    flags << 'customer_support' if customer_support?(text)
+    flags << 'download_spam' if download_links?(text, urls, protip.title)
+    flags << 'recognized_format' if recognized_format?(text)
+    flags << 'mostly_url' if mostly_url?(text, urls)
+    flags << 'weird_characters' if weird_characters?(text)
+
+    flags
+  end
+
+  def user_flags(user)
+    flags = []
+    text = [user.title, user.username, user.about].flatten.join("\n")
+    urls = URI.extract(text).compact
+
+    flags << 'bad_links' if bad_links?(text, urls)
+    flags << 'customer_support' if customer_support?(text)
+    flags << 'download_spam' if download_links?(text, urls, user.username)
+    flags << 'recognized_format' if recognized_format?(text)
+    flags << 'many_spaces' if many_spaces?(text, urls, user.username)
+    flags << 'mostly_url' if mostly_url?(text, urls)
+    flags << 'weird_characters' if weird_characters?(text)
+
+    flags
+  end
+
+end
+
diff --git a/app/services/smyte.rb b/app/services/smyte.rb
diff --git a/lib/tasks/spam.rake b/lib/tasks/spam.rake
@@ -0,0 +1,39 @@
+namespace :spam do
+  task :sweep => :environment do
+    protips = Protip.where('created_at > ?', 7.days.ago).where(bad_content: false)
+    good = []
+    protips.each do |p|
+      flags = Spaminator.new.protip_flags(p)
+      if flags.any?
+        puts "#{p.id} – #{p.title} – #{p.body[0..100].gsub("\n", '')}"
+        puts "#{flags.inspect}" if flags.any?
+        puts
+
+        p.bad_content = true
+        p.user.bad_user = true
+        p.save
+      else
+        good << p
+      end
+    end
+
+    users = User.where('created_at > ?', 7.days.ago).where(bad_user: false)
+    users.map do |u|
+      flags = Spaminator.new.user_flags(u)
+      if flags.any?
+        puts "#{u.id} – #{u.username} – #{(u.about || '')[0..100].gsub("\n", '')}"
+        puts "#{flags.inspect}" if flags.any?
+        puts
+
+        u.bad_user!
+      else
+        good << u
+      end
+    end
+
+    puts "Good"
+    good.each do |e|
+      puts "#{e.class}:#{e.id} – #{e.try(:username) || e.title}"
+    end
+  end
+end
