Collect projects Value Stream stages

app/models/analytics/cycle_analytics/project_stage.rb

@@ -26,6 +26,12 @@ module Analytics
         :project_id
       end
 
+      def self.distinct_stages_within_hierarchy(group)
+        with_preloaded_labels
+          .where(project_id: group.all_projects.select(:id))
+          .select("DISTINCT ON(stage_event_hash_id) #{quoted_table_name}.*")
+      end
+
       private
 
       # Project should belong to a group when the stage has Label based events since only GroupLabels are allowed.

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

@@ -29,6 +29,7 @@ module Gitlab
         def stages
           [
             *group_stages,
+            *project_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)
@@ -42,6 +43,10 @@ module Gitlab
           @group_stages ||= ::Analytics::CycleAnalytics::GroupStage.distinct_stages_within_hierarchy(group)
         end
 
+        def project_stages
+          @project_stages ||= ::Analytics::CycleAnalytics::ProjectStage.distinct_stages_within_hierarchy(group)
+        end
+
         def in_memory_lead_time_stage
           ::Analytics::CycleAnalytics::GroupStage.new(
             name: 'lead time', # not visible to the user

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

@@ -5,9 +5,13 @@ 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) }
+  let_it_be(:common_stage_params) { { start_event_identifier: :issue_created, end_event_identifier: :issue_first_associated_with_milestone } }
+  let_it_be(:stage_2) { create(:cycle_analytics_group_stage, group: group, **common_stage_params) }
+  let_it_be(:stage_duplicate) { create(:cycle_analytics_group_stage, group: group, **common_stage_params) }
+  let_it_be(:stage_triplicate) { create(:cycle_analytics_group_stage, group: group, **common_stage_params) }
+
+  let_it_be(:project) { create(:project, group: group) }
+  let_it_be(:project_stage) { create(:cycle_analytics_project_stage, project: project, **common_stage_params) }
 
   subject(:distinct_stages) { described_class.new(group: group).stages }
 

spec/models/analytics/cycle_analytics/project_stage_spec.rb

@@ -29,4 +29,29 @@ RSpec.describe Analytics::CycleAnalytics::ProjectStage do
       let(:default_params) { { project: project } }
     end
   end
+
+  describe '.distinct_stages_within_hierarchy' do
+    let_it_be(:top_level_group) { create(:group) }
+    let_it_be(:sub_group_1) { create(:group, parent: top_level_group) }
+    let_it_be(:sub_group_2) { create(:group, parent: sub_group_1) }
+
+    let_it_be(:project_1) { create(:project, group: sub_group_1) }
+    let_it_be(:project_2) { create(:project, group: sub_group_2) }
+    let_it_be(:project_3) { create(:project, group: top_level_group) }
+
+    let_it_be(:stage1) { create(:cycle_analytics_project_stage, project: project_1, start_event_identifier: :issue_created, end_event_identifier: :issue_deployed_to_production) }
+    let_it_be(:stage2) { create(:cycle_analytics_project_stage, project: project_3, start_event_identifier: :issue_created, end_event_identifier: :issue_deployed_to_production) }
+
+    let_it_be(:stage3) { create(:cycle_analytics_project_stage, project: project_1, start_event_identifier: :merge_request_created, end_event_identifier: :merge_request_merged) }
+    let_it_be(:stage4) { create(:cycle_analytics_project_stage, project: project_3, start_event_identifier: :merge_request_created, end_event_identifier: :merge_request_merged) }
+
+    subject(:distinct_start_and_end_event_identifiers) { described_class.distinct_stages_within_hierarchy(top_level_group).to_a.pluck(:start_event_identifier, :end_event_identifier) }
+
+    it 'returns distinct stages by start and end events (using stage_event_hash_id)' do
+      expect(distinct_start_and_end_event_identifiers).to match_array([
+        %w[issue_created issue_deployed_to_production],
+        %w[merge_request_created merge_request_merged]
+      ])
+    end
+  end
 end