Delete the set_cluster method definition and calls from Alegre Bot

For FactsFirstPH, elections project we run in the Philippines in early 2022, we implemented the very first version of shared feeds and global clusters. Part of that logic included adding new items to clusters automatically. We’re changing this logic, so before proceeding with the new logic, it’s important to get rid of the previous one.

app/models/bot/alegre.rb

@@ -162,7 +162,6 @@ def self.run(body)
         self.get_extracted_text(pm)
         self.get_flags(pm)
         self.auto_transcription(pm)
-        self.set_cluster(pm)
         handled = true
       end
     rescue StandardError => e
@@ -175,29 +174,6 @@ def self.run(body)
     handled
   end
 
-  def self.set_cluster(pm, force = false)
-    pm = ProjectMedia.find(pm.id)
-    return if (!pm.cluster_id.blank? || !ProjectMedia.where(team_id: pm.team_id).where.not(cluster_id: nil).exists?) && !force
-    team_ids = ProjectMedia.where.not(cluster_id: nil).group(:team_id).count.keys
-    thresholds = {
-      audio: { value: CheckConfig.get('audio_cluster_similarity_threshold', 0.8, :float) },
-      video: { value: CheckConfig.get('video_cluster_similarity_threshold', 0.8, :float) },
-      image: { value: CheckConfig.get('image_cluster_similarity_threshold', 0.9, :float) },
-      text: { value: CheckConfig.get('text_cluster_similarity_threshold', 0.9, :float) }
-    }
-    ids_and_scores = pm.similar_items_ids_and_scores(team_ids, thresholds)
-    main_id = ids_and_scores.max_by{ |_pm_id, score_and_context| score_and_context[:score] }&.first
-    main = ProjectMedia.find_by_id(main_id.to_i)
-    cluster = main&.cluster
-    unless cluster
-      cluster = Cluster.new
-      cluster.project_media = pm
-      cluster.skip_check_ability = true
-      cluster.save!
-    end
-    cluster.project_medias << pm
-    cluster
-  end
 
   def self.get_number_of_words(text)
     # Get the number of space-separated words (Does not work with Chinese/Japanese)

app/models/concerns/relationship_bulk.rb

@@ -73,7 +73,7 @@ def run_update_callbacks(ids_json, extra_options_json)
       index_alias = CheckElasticSearchModel.get_index_alias
       es_body = []
       versions = []
-      callbacks = [:reset_counters, :update_counters, :set_cluster, :propagate_inversion]
+      callbacks = [:reset_counters, :update_counters, :propagate_inversion]
       target_ids = []
       Relationship.where(id: ids, source_id: extra_options['source_id']).find_each do |r|
         target_ids << r.target_id

app/models/relationship.rb

@@ -12,7 +12,6 @@ class Relationship < ApplicationRecord
 
   before_validation :set_user, on: :create
   before_validation :set_confirmed, if: :is_being_confirmed?, on: :update
-  before_validation :set_cluster, if: :is_being_confirmed?, on: :update
   validate :relationship_type_is_valid, :items_are_from_the_same_team
   validate :target_not_published_report, on: :create
   validate :similar_item_exists, on: :create, if: proc { |r| r.is_suggested? }
@@ -31,7 +30,6 @@ class Relationship < ApplicationRecord
   after_destroy :turn_on_unmatched_field, if: proc { |r| r.is_confirmed? || r.is_suggested? }
   after_commit :update_counter_and_elasticsearch, on: [:create, :update]
   after_commit :update_counters, :destroy_elasticsearch_relation, on: :destroy
-  after_commit :set_cluster, on: [:create]
 
   has_paper_trail on: [:create, :update, :destroy], if: proc { |x| User.current.present? && !x.is_being_copied? }, versions: { class_name: 'Version' }
 
@@ -282,21 +280,6 @@ def set_confirmed
     end
   end
 
-  def set_cluster
-    if self.relationship_type.to_json == Relationship.confirmed_type.to_json && User.current && User.current&.id != BotUser.alegre_user&.id
-      pm = self.target
-      new_cluster = self.source.cluster
-      old_cluster = pm.cluster
-      if old_cluster.nil? || (old_cluster.size == 1 && old_cluster.project_media_id == pm.id)
-        unless old_cluster.nil?
-          old_cluster.skip_check_ability = true
-          old_cluster.destroy!
-        end
-        new_cluster.project_medias << pm unless new_cluster.nil?
-      end
-    end
-  end
-
   def turn_off_unmatched_field
     set_unmatched_field(0)
   end

lib/tasks/check_clusters.rake

@@ -138,11 +138,6 @@ namespace :check do
       team = Team.find_by_slug(slug)
       n = ProjectMedia.where(team_id: team.id).count
       i = 0
-      ProjectMedia.where(team_id: team.id).order('id ASC').find_each do |pm|
-        i += 1
-        c = Bot::Alegre.set_cluster(pm, true)
-        log "[#{i}/#{n}] [#{Time.now}] Adding item #{pm.id} to the clusters... added to cluster #{c.id}"
-      end
     end
   end
 end

test/models/bot/alegre_2_test.rb

@@ -673,14 +673,10 @@ def teardown
 
     ProjectMedia.any_instance.stubs(:similar_items_ids_and_scores).returns({ pm1.id => { score: 0.9 }, pm2.id => { score: 0.8 } })
     pm3 = create_project_media team: @team
-    Bot::Alegre.set_cluster(pm3)
     assert_equal c1.id, pm3.reload.cluster_id
 
     ProjectMedia.any_instance.stubs(:similar_items_ids_and_scores).returns({})
     pm4 = create_project_media team: @team
-    assert_difference 'Cluster.count' do
-      Bot::Alegre.set_cluster(pm4)
-    end
 
     ProjectMedia.any_instance.unstub(:similar_items_ids_and_scores)
   end

test/models/cluster_test.rb

@@ -87,17 +87,6 @@ def setup
     end
   end
 
-  test "should set cluster" do
-    t = create_team
-    pm1 = create_project_media team: t
-    c = create_cluster
-    c.project_medias << pm1
-    pm2 = create_project_media team: t
-    ProjectMedia.any_instance.stubs(:similar_items_ids_and_scores).returns({ pm1.id => { score: 0.9, context: {} }, random_number => { score: 0.8, context: { foo: 'bar' } } })
-    assert_equal c, Bot::Alegre.set_cluster(pm2)
-    ProjectMedia.any_instance.unstub(:similar_items_ids_and_scores)
-  end
-
   test "should get requests count" do
     create_annotation_type_and_fields('Smooch', { 'Data' => ['JSON', true] })
     RequestStore.store[:skip_cached_field_update] = false