Collect all distinct VSA stages

ee/app/services/analytics/cycle_analytics/data_loader_service.rb

@@ -173,8 +173,9 @@ module Analytics
       end
 
       def stages
-        @stages ||= Analytics::CycleAnalytics::GroupStage
-          .distinct_stages_within_hierarchy(group)
+        @stages ||= Gitlab::Analytics::CycleAnalytics::DistinctStageLoader
+          .new(group: group)
+          .stages
           .select { |stage| stage.start_event.object_type == model }
       end
 

ee/lib/gitlab/analytics/cycle_analytics/distinct_stage_loader.rb

+# frozen_string_literal: true
+
+module Gitlab
+  module Analytics
+    module CycleAnalytics
+      # This class is responsible for loading distinct stage records
+      # within the group hierarchy. The stage metadata is uniquely identified
+      # by its stage_event_hash_id column. There can be multiple stages with the
+      # same metadata (created with different names or within different subgroups).
+      #
+      # Example:
+      #
+      # Stage 1 (config: issue_created - issue_closed)
+      # Stage 2 (config: issue_created - issue_deployed_to_production)
+      # Stage 3 (config: issue_created - issue_closed) # same metadata as #1
+      #
+      # Expected results:
+      #
+      # Stage 1, Stage 2
+      #
+      # The class also adds two "in-memory" stages into the distinct calculation which
+      # represent the CycleTime and LeadTime metrics. These metrics are calculated as
+      # pre-defined VSA stages.
+      class DistinctStageLoader
+        def initialize(group:)
+          @group = group
+        end
+
+        def stages
+          [
+            *group_stages,
+            add_stage_event_hash_id(in_memory_lead_time_stage),
+            add_stage_event_hash_id(in_memory_cycle_time_stage)
+          ].uniq(&:stage_event_hash_id)
+        end
+
+        private
+
+        attr_reader :group
+
+        def group_stages
+          @group_stages ||= ::Analytics::CycleAnalytics::GroupStage.distinct_stages_within_hierarchy(group)
+        end
+
+        def in_memory_lead_time_stage
+          ::Analytics::CycleAnalytics::GroupStage.new(
+            name: 'lead time', # not visible to the user
+            start_event_identifier: Summary::LeadTime.start_event_identifier,
+            end_event_identifier: Summary::LeadTime.end_event_identifier,
+            group: group
+          )
+        end
+
+        def in_memory_cycle_time_stage
+          ::Analytics::CycleAnalytics::GroupStage.new(
+            name: 'cycle time',
+            start_event_identifier: Summary::CycleTime.start_event_identifier,
+            end_event_identifier: Summary::CycleTime.end_event_identifier,
+            group: group
+          )
+        end
+
+        # rubocop: disable CodeReuse/ActiveRecord
+        def add_stage_event_hash_id(stage)
+          # find or create the stage event hash
+          hash_record = ::Analytics::CycleAnalytics::StageEventHash.find_by(hash_sha256: stage.events_hash_code)
+          stage.stage_event_hash_id = if hash_record
+                                        hash_record.id
+                                      else
+                                        ::Analytics::CycleAnalytics::StageEventHash.record_id_by_hash_sha256(stage.events_hash_code)
+                                      end
+
+          stage
+        end
+        # rubocop: enable CodeReuse/ActiveRecord
+      end
+    end
+  end
+end

ee/lib/gitlab/analytics/cycle_analytics/summary/base_time.rb

@@ -21,11 +21,11 @@ module Gitlab
             n_('day', 'days', value)
           end
 
-          def start_event_identifier
+          def self.start_event_identifier
             raise NotImplementedError, "Expected #{self.name} to implement start_event_identifier"
           end
 
-          def end_event_identifier
+          def self.end_event_identifier
             raise NotImplementedError, "Expected #{self.name} to implement end_event_identifier"
           end
 
@@ -36,8 +36,8 @@ module Gitlab
           private
 
           def assign_event_identifiers
-            @stage.start_event_identifier = start_event_identifier
-            @stage.end_event_identifier = end_event_identifier
+            @stage.start_event_identifier = self.class.start_event_identifier
+            @stage.end_event_identifier = self.class.end_event_identifier
           end
 
           def data_collector

ee/lib/gitlab/analytics/cycle_analytics/summary/cycle_time.rb

@@ -9,11 +9,11 @@ module Gitlab
             _('Cycle Time')
           end
 
-          def start_event_identifier
+          def self.start_event_identifier
             :issue_first_mentioned_in_commit
           end
 
-          def end_event_identifier
+          def self.end_event_identifier
             :issue_closed
           end
         end

ee/lib/gitlab/analytics/cycle_analytics/summary/lead_time.rb

@@ -9,11 +9,11 @@ module Gitlab
             _('Lead Time')
           end
 
-          def start_event_identifier
+          def self.start_event_identifier
             :issue_created
           end
 
-          def end_event_identifier
+          def self.end_event_identifier
             :issue_closed
           end
 

ee/spec/lib/gitlab/analytics/cycle_analytics/distinct_stage_loader_spec.rb

+# frozen_string_literal: true
+
+require 'spec_helper'
+
+RSpec.describe Gitlab::Analytics::CycleAnalytics::DistinctStageLoader do
+  let_it_be(:group) { create(:group) }
+  let_it_be(:stage_1) { create(:cycle_analytics_group_stage, group: group, start_event_identifier: :merge_request_created, end_event_identifier: :merge_request_merged) }
+  let_it_be(:stage_2) { create(:cycle_analytics_group_stage, group: group, start_event_identifier: :issue_created, end_event_identifier: :issue_first_associated_with_milestone) }
+  let_it_be(:stage_duplicate) { create(:cycle_analytics_group_stage, group: group, start_event_identifier: :issue_created, end_event_identifier: :issue_first_associated_with_milestone) }
+  let_it_be(:stage_triplicate) { create(:cycle_analytics_group_stage, group: group, start_event_identifier: :issue_created, end_event_identifier: :issue_first_associated_with_milestone) }
+
+  subject(:distinct_stages) { described_class.new(group: group).stages }
+
+  it 'returns the distinct stages by stage_event_hash_id' do
+    distinct_stage_hash_ids = subject.map(&:stage_event_hash_id)
+
+    expect(distinct_stage_hash_ids).to eq(distinct_stage_hash_ids.uniq)
+  end
+
+  context 'when lead time and cycle time are not defined as stages' do
+    it 'returns in-memory stages' do
+      lead_time = distinct_stages.find { |stage| stage.name == 'lead time' }
+      cycle_time = distinct_stages.find { |stage| stage.name == 'cycle time' }
+
+      expect(lead_time).to be_present
+      expect(cycle_time).to be_present
+      expect(lead_time.stage_event_hash_id).not_to be_nil
+      expect(cycle_time.stage_event_hash_id).not_to be_nil
+
+      expect(lead_time.stage_event_hash_id).not_to eq(cycle_time.stage_event_hash_id)
+    end
+
+    it 'creates two stage event hash records' do
+      expect { distinct_stages }.to change { Analytics::CycleAnalytics::StageEventHash.count }.by(2)
+    end
+
+    it 'returns 4 stages' do
+      expect(distinct_stages.size).to eq(4)
+    end
+  end
+
+  context 'when lead time and cycle time are persisted stages' do
+    let_it_be(:cycle_time) do
+      create(:cycle_analytics_group_stage,
+             group: group,
+             start_event_identifier: :issue_created,
+             end_event_identifier: :issue_first_associated_with_milestone)
+    end
+
+    let_it_be(:lead_tiem) do
+      create(:cycle_analytics_group_stage,
+             group: group,
+             start_event_identifier: :issue_created,
+             end_event_identifier: :issue_first_associated_with_milestone)
+    end
+
+    it 'does not create extra stage event hash records' do
+      expect { distinct_stages }.to change { Analytics::CycleAnalytics::StageEventHash.count }
+    end
+  end
+end