Add latest changes from gitlab-org/gitlab@13-3-stable-ee

4 files changed, 243 insertions, 20 deletions

diff --git a/lib/gitlab/usage_data_counters/hll_redis_counter.rb b/lib/gitlab/usage_data_counters/hll_redis_counter.rb
new file mode 100644
index 00000000000..c9c39225068
--- /dev/null
+++ b/lib/gitlab/usage_data_counters/hll_redis_counter.rb
@@ -0,0 +1,149 @@
+# frozen_string_literal: true
+
+module Gitlab
+  module UsageDataCounters
+    module HLLRedisCounter
+      DEFAULT_WEEKLY_KEY_EXPIRY_LENGTH = 6.weeks
+      DEFAULT_DAILY_KEY_EXPIRY_LENGTH = 29.days
+      DEFAULT_REDIS_SLOT = ''.freeze
+
+      UnknownEvent = Class.new(StandardError)
+      UnknownAggregation = Class.new(StandardError)
+
+      KNOWN_EVENTS_PATH = 'lib/gitlab/usage_data_counters/known_events.yml'.freeze
+      ALLOWED_AGGREGATIONS = %i(daily weekly).freeze
+
+      # Track event on entity_id
+      # Increment a Redis HLL counter for unique event_name and entity_id
+      #
+      # All events should be added to know_events file lib/gitlab/usage_data_counters/known_events.yml
+      #
+      # Event example:
+      #
+      # - name: g_compliance_dashboard # Unique event name
+      #   redis_slot: compliance       # Optional slot name, if not defined it will use name as a slot, used for totals
+      #   category: compliance         # Group events in categories
+      #   expiry: 29                   # Optional expiration time in days, default value 29 days for daily and 6.weeks for weekly
+      #   aggregation: daily           # Aggregation level, keys are stored daily or weekly
+      #
+      # Usage:
+      #
+      # * Track event: Gitlab::UsageDataCounters::HLLRedisCounter.track_event(user_id, 'g_compliance_dashboard')
+      # * Get unique counts per user: Gitlab::UsageDataCounters::HLLRedisCounter.unique_events(event_names: 'g_compliance_dashboard', start_date: 28.days.ago, end_date: Date.current)
+      class << self
+        def track_event(entity_id, event_name, time = Time.zone.now)
+          event = event_for(event_name)
+
+          raise UnknownEvent.new("Unknown event #{event_name}") unless event.present?
+
+          Gitlab::Redis::HLL.add(key: redis_key(event, time), value: entity_id, expiry: expiry(event))
+        end
+
+        def unique_events(event_names:, start_date:, end_date:)
+          events = events_for(Array(event_names))
+
+          raise 'Events should be in same slot' unless events_in_same_slot?(events)
+          raise 'Events should be in same category' unless events_in_same_category?(events)
+          raise 'Events should have same aggregation level' unless events_same_aggregation?(events)
+
+          aggregation = events.first[:aggregation]
+
+          keys = keys_for_aggregation(aggregation, events: events, start_date: start_date, end_date: end_date)
+
+          Gitlab::Redis::HLL.count(keys: keys)
+        end
+
+        def events_for_category(category)
+          known_events.select { |event| event[:category] == category }.map { |event| event[:name] }
+        end
+
+        private
+
+        def keys_for_aggregation(aggregation, events:, start_date:, end_date:)
+          if aggregation.to_sym == :daily
+            daily_redis_keys(events: events, start_date: start_date, end_date: end_date)
+          else
+            weekly_redis_keys(events: events, start_date: start_date, end_date: end_date)
+          end
+        end
+
+        def known_events
+          @known_events ||= YAML.load_file(Rails.root.join(KNOWN_EVENTS_PATH)).map(&:with_indifferent_access)
+        end
+
+        def known_events_names
+          known_events.map { |event| event[:name] }
+        end
+
+        def events_in_same_slot?(events)
+          slot = events.first[:redis_slot]
+          events.all? { |event| event[:redis_slot] == slot }
+        end
+
+        def events_in_same_category?(events)
+          category = events.first[:category]
+          events.all? { |event| event[:category] == category }
+        end
+
+        def events_same_aggregation?(events)
+          aggregation = events.first[:aggregation]
+          events.all? { |event| event[:aggregation] == aggregation }
+        end
+
+        def expiry(event)
+          return event[:expiry] if event[:expiry].present?
+
+          event[:aggregation].to_sym == :daily ? DEFAULT_DAILY_KEY_EXPIRY_LENGTH : DEFAULT_WEEKLY_KEY_EXPIRY_LENGTH
+        end
+
+        def event_for(event_name)
+          known_events.find { |event| event[:name] == event_name }
+        end
+
+        def events_for(event_names)
+          known_events.select { |event| event_names.include?(event[:name]) }
+        end
+
+        def redis_slot(event)
+          event[:redis_slot] || DEFAULT_REDIS_SLOT
+        end
+
+        # Compose the key in order to store events daily or weekly
+        def redis_key(event, time)
+          raise UnknownEvent.new("Unknown event #{event[:name]}") unless known_events_names.include?(event[:name].to_s)
+          raise UnknownAggregation.new("Use :daily or :weekly aggregation") unless ALLOWED_AGGREGATIONS.include?(event[:aggregation].to_sym)
+
+          slot = redis_slot(event)
+          key = if slot.present?
+                  event[:name].to_s.gsub(slot, "{#{slot}}")
+                else
+                  "{#{event[:name]}}"
+                end
+
+          if event[:aggregation].to_sym == :daily
+            year_day = time.strftime('%G-%j')
+            "#{year_day}-#{key}"
+          else
+            year_week = time.strftime('%G-%V')
+            "#{key}-#{year_week}"
+          end
+        end
+
+        def daily_redis_keys(events:, start_date:, end_date:)
+          (start_date.to_date..end_date.to_date).map do |date|
+            events.map { |event| redis_key(event, date) }
+          end.flatten
+        end
+
+        def weekly_redis_keys(events:, start_date:, end_date:)
+          weeks = end_date.to_date.cweek - start_date.to_date.cweek
+          weeks = 1 if weeks == 0
+
+          (0..(weeks - 1)).map do |week_increment|
+            events.map { |event| redis_key(event, start_date + week_increment * 7.days) }
+          end.flatten
+        end
+      end
+    end
+  end
+end
diff --git a/lib/gitlab/usage_data_counters/known_events.yml b/lib/gitlab/usage_data_counters/known_events.yml
new file mode 100644
index 00000000000..b7e516fa8b1
--- /dev/null
+++ b/lib/gitlab/usage_data_counters/known_events.yml
@@ -0,0 +1,88 @@
+---
+# Compliance category
+- name: g_compliance_dashboard
+  redis_slot: compliance
+  category: compliance
+  expiry: 84 # expiration time in days, equivalent to 12 weeks
+  aggregation: weekly
+- name: g_compliance_audit_events
+  category: compliance
+  redis_slot: compliance
+  expiry: 84
+  aggregation: weekly
+- name: i_compliance_audit_events
+  category: compliance
+  redis_slot: compliance
+  expiry: 84
+  aggregation: weekly
+- name: i_compliance_credential_inventory
+  category: compliance
+  redis_slot: compliance
+  expiry: 84
+  aggregation: weekly
+# Analytics category
+- name: g_analytics_contribution
+  category: analytics
+  redis_slot: analytics
+  expiry: 84
+  aggregation: weekly
+- name: g_analytics_insights
+  category: analytics
+  redis_slot: analytics
+  expiry: 84
+  aggregation: weekly
+- name: g_analytics_issues
+  category: analytics
+  redis_slot: analytics
+  expiry: 84
+  aggregation: weekly
+- name: g_analytics_productivity
+  category: analytics
+  redis_slot: analytics
+  expiry: 84
+  aggregation: weekly
+- name: g_analytics_valuestream
+  category: analytics
+  redis_slot: analytics
+  expiry: 84
+  aggregation: weekly
+- name: p_analytics_pipelines
+  category: analytics
+  redis_slot: analytics
+  expiry: 84
+  aggregation: weekly
+- name: p_analytics_code_reviews
+  category: analytics
+  redis_slot: analytics
+  expiry: 84
+  aggregation: weekly
+- name: p_analytics_valuestream
+  category: analytics
+  redis_slot: analytics
+  expiry: 84
+  aggregation: weekly
+- name: p_analytics_insights
+  category: analytics
+  redis_slot: analytics
+  expiry: 84
+  aggregation: weekly
+- name: p_analytics_issues
+  category: analytics
+  redis_slot: analytics
+  expiry: 84
+  aggregation: weekly
+- name: p_analytics_repo
+  category: analytics
+  redis_slot: analytics
+  expiry: 84
+  aggregation: weekly
+- name: i_analytics_cohorts
+  category: analytics
+  redis_slot: analytics
+  expiry: 84
+  aggregation: weekly
+- name: i_analytics_dev_ops_score
+  category: analytics
+  redis_slot: analytics
+  expiry: 84
+  aggregation: weekly
diff --git a/lib/gitlab/usage_data_counters/track_unique_actions.rb b/lib/gitlab/usage_data_counters/track_unique_actions.rb
index 9fb5a29748e..0df982572a4 100644
--- a/lib/gitlab/usage_data_counters/track_unique_actions.rb
+++ b/lib/gitlab/usage_data_counters/track_unique_actions.rb
@@ -4,7 +4,6 @@ module Gitlab
   module UsageDataCounters
     module TrackUniqueActions
       KEY_EXPIRY_LENGTH = 29.days
-      FEATURE_FLAG = :track_unique_actions
 
       WIKI_ACTION = :wiki_action
       DESIGN_ACTION = :design_action
@@ -27,24 +26,22 @@ module Gitlab
       }).freeze
 
       class << self
-        def track_action(event_action:, event_target:, author_id:, time: Time.zone.now)
+        def track_event(event_action:, event_target:, author_id:, time: Time.zone.now)
           return unless Gitlab::CurrentSettings.usage_ping_enabled
-          return unless Feature.enabled?(FEATURE_FLAG)
           return unless valid_target?(event_target)
           return unless valid_action?(event_action)
 
           transformed_target = transform_target(event_target)
           transformed_action = transform_action(event_action, transformed_target)
+          target_key = key(transformed_action, time)
 
-          add_event(transformed_action, author_id, time)
+          Gitlab::Redis::HLL.add(key: target_key, value: author_id, expiry: KEY_EXPIRY_LENGTH)
         end
 
-        def count_unique_events(event_action:, date_from:, date_to:)
+        def count_unique(event_action:, date_from:, date_to:)
           keys = (date_from.to_date..date_to.to_date).map { |date| key(event_action, date) }
 
-          Gitlab::Redis::SharedState.with do |redis|
-            redis.pfcount(*keys)
-          end
+          Gitlab::Redis::HLL.count(keys: keys)
         end
 
         private
@@ -69,17 +66,6 @@ module Gitlab
           year_day = date.strftime('%G-%j')
           "#{year_day}-{#{event_action}}"
         end
-
-        def add_event(event_action, author_id, date)
-          target_key = key(event_action, date)
-
-          Gitlab::Redis::SharedState.with do |redis|
-            redis.multi do |multi|
-              multi.pfadd(target_key, author_id)
-              multi.expire(target_key, KEY_EXPIRY_LENGTH)
-            end
-          end
-        end
       end
     end
   end
diff --git a/lib/gitlab/usage_data_counters/wiki_page_counter.rb b/lib/gitlab/usage_data_counters/wiki_page_counter.rb
index 9cfe0be5bab..6c3fe842344 100644
--- a/lib/gitlab/usage_data_counters/wiki_page_counter.rb
+++ b/lib/gitlab/usage_data_counters/wiki_page_counter.rb
@@ -2,7 +2,7 @@
 
 module Gitlab::UsageDataCounters
   class WikiPageCounter < BaseCounter
-    KNOWN_EVENTS = %w[create update delete].freeze
+    KNOWN_EVENTS = %w[view create update delete].freeze
     PREFIX = 'wiki_pages'
   end
 end