Refactor anomaly detection task

lib/tasks/detect_exercise_anomalies.rake

@@ -1,19 +1,36 @@
 namespace :detect_exercise_anomalies do
 
-  task :with_at_least, [:number_of_exercises, :number_of_solutions] => :environment do |task, args|
-    number_of_exercises = args.number_of_exercises
-    number_of_solutions = args.number_of_solutions
-
-    puts "\tSearching for exercise collections with at least #{number_of_exercises} exercises and #{number_of_solutions} users."
-
   # These factors determine if an exercise is an anomaly, given the average working time (avg):
   # (avg * MIN_TIME_FACTOR) <= working_time <= (avg * MAX_TIME_FACTOR)
   MIN_TIME_FACTOR = 0.1
   MAX_TIME_FACTOR = 2
 
+  task :with_at_least, [:number_of_exercises, :number_of_solutions] => :environment do |task, args|
+    number_of_exercises = args[:number_of_exercises]
+    number_of_solutions = args[:number_of_solutions]
+
+    puts "Searching for exercise collections with at least #{number_of_exercises} exercises and #{number_of_solutions} users."
     # Get all exercise collections that have at least the specified amount of exercises and at least the specified
     # number of submissions AND are flagged for anomaly detection
-    collections = ExerciseCollection
+    collections = get_collections(number_of_exercises, number_of_solutions)
+    puts "Found #{collections.length}."
+
+    collections.each do |collection|
+      puts "\t- #{collection}"
+      anomalies = find_anomalies(collection)
+
+      if anomalies.length > 0 and not collection.user.nil?
+        puts "\t\tAnomalies: #{anomalies}\n"
+        notify_collection_author(collection, anomalies)
+        notify_users(collection, anomalies)
+        reset_anomaly_detection_flag(collection)
+      end
+    end
+    puts 'Done.'
+  end
+
+  def get_collections(number_of_exercises, number_of_solutions)
+    ExerciseCollection
       .where(:use_anomaly_detection => true)
       .joins("join exercise_collections_exercises ece on exercise_collections.id = ece.exercise_collection_id
                             join
@@ -25,11 +42,9 @@ namespace :detect_exercise_anomalies do
                               ) as exercises_with_submissions on exercises_with_submissions.id = ece.exercise_id")
       .group('exercise_collections.id')
       .having('count(exercises_with_submissions.id) > ?', number_of_exercises)
+  end
 
-    puts "\tFound #{collections.length}."
-
-    collections.each do |collection|
-      puts "\t- #{collection}"
+  def find_anomalies(collection)
     working_times = {}
     collection.exercises.each do |exercise|
       puts "\t\t> #{exercise.title}"
@@ -38,16 +53,17 @@ namespace :detect_exercise_anomalies do
       working_times[exercise.id] = seconds
     end
     average = working_times.values.reduce(:+) / working_times.size
-      anomalies = working_times.select do |exercise_id, working_time|
+    working_times.select do |exercise_id, working_time|
       working_time > average * MAX_TIME_FACTOR or working_time < average * MIN_TIME_FACTOR
     end
+  end
 
-      if anomalies.length > 0 and not collection.user.nil?
-        puts "\t\tAnomalies: #{anomalies}\n"
-
+  def notify_collection_author(collection, anomalies)
     puts "\t\tSending E-Mail to author (#{collection.user.displayname} <#{collection.user.email}>)..."
     UserMailer.exercise_anomaly_detected(collection, anomalies).deliver_now
+  end
 
+  def notify_users(collection, anomalies)
     puts "\t\tSending E-Mails to best and worst performing users of each anomaly..."
     anomalies.each do |exercise_id, average_working_time|
       submissions = Submission.find_by_sql(['
@@ -73,13 +89,12 @@ namespace :detect_exercise_anomalies do
       puts "\t\t\tbest performers: #{best_performers}"
       puts "\t\t\tworst performers: #{worst_performers}"
     end
+  end
 
+  def reset_anomaly_detection_flag(collection)
     puts "\t\tResetting flag..."
     collection.use_anomaly_detection = false
     collection.save!
   end
-    end
-    puts "\tDone."
-  end
 
 end