Rake to dump list of screenshots for future ML research

Rakefile

@@ -582,5 +582,18 @@ task :generate_sitemap => [:environment] do
     gz.write %{</sitemapindex>}
   end
 
+  desc 'ml_screenshots_list_dump'
+  task :ml_screenshots_list_dump => [:environment] do
+    ['phishing', 'spam', 'ham', nil].each do |classifier|
+      File.open("./files/screenshot-urls-#{classifier.to_s}.txt", 'w') do |fp|
+        SiteFile.where(classifier: classifier).where(path: 'index.html').each do |site_file|
+          begin
+            fp.write "#{site_file.site.screenshot_url('index.html', Site::SCREENSHOT_RESOLUTIONS.first)}\n"
+          rescue NoMethodError
+          end
+        end
+      end
+    end
+  end
 
 end