Assets investigation - data extraction rake tasks

lib/tasks/attachments/check_assets.rake

@@ -0,0 +1,24 @@
+desc "Check assets"
+task check_assets: :environment do
+  file = File.open("./lib/tasks/attachments/attachment_data_deleted_in_wh_replaced_in_both_but_replacements_are_draft.txt", "a")
+
+  File.readlines("./lib/tasks/attachments/ads_to_check.txt", chomp: true).each do |line|
+    attachment_data_id, asset_manager_id, = line.split(",")
+
+    begin
+      asset_manager_response = GdsApi.asset_manager.asset(asset_manager_id).to_h
+      rep_response = GdsApi.asset_manager.asset(asset_manager_response["replacement_id"]).to_h
+    rescue GdsApi::HTTPNotFound
+      next
+    end
+
+    if rep_response["draft"]
+      puts "ad: #{attachment_data_id}, am_id: #{asset_manager_id}, rep_id: #{asset_manager_response['replacement_id']}, rep_del: #{rep_response['deleted']}, rep_rep: #{!rep_response['replacement_id'].nil?}"
+      file << "#{attachment_data_id},#{asset_manager_id},#{asset_manager_response['replacement_id']},#{rep_response['deleted']},#{!rep_response['replacement_id'].nil?}" << "\n"
+    else
+      print "."
+    end
+  end
+
+  file.close
+end

lib/tasks/attachments/check_attachables.rake

@@ -0,0 +1,19 @@
+desc "Check attachables"
+task check_attachables: :environment do
+  File.readlines("./lib/tasks/attachments/ads_to_check.txt", chomp: true).each do |line|
+    attachment_data_id = line.split(",").first
+    states = AttachmentData.find(attachment_data_id).attachments.map(&:attachable).map { |a| a&.state }
+    if !valid_state(states)
+      puts "NOT OK: #{attachment_data_id}: [#{states.join(', ')}]"
+    else
+      print "."
+    end
+  end
+end
+
+def valid_state(states)
+  return false unless states[-1].nil?
+  return false unless states.reject { |state| ["superseded", nil].include? state }.empty?
+
+  true
+end

lib/tasks/attachments/find_all_superseded_not_deleted_attachments.rake

@@ -0,0 +1,67 @@
+desc "Report on AttachmentData in WH that is not deleted"
+task attachment_data_not_deleted: :environment do
+  file1 = File.open("./lib/tasks/attachments/ad_not_deleted_but_replaced_OK.txt", "a")
+  file11 = File.open("./lib/tasks/attachments/ad_not_deleted_but_replaced_with_draft_replacement.txt", "a")
+  file2 = File.open("./lib/tasks/attachments/ad_not_deleted_replaced_in_wh_but_not_in_am.txt", "a")
+  file3 = File.open("./lib/tasks/attachments/ad_not_deleted_not_replaced_in_wh.txt", "a")
+
+  AttachmentData.find_each.map do |attachment_data|
+    attachments = attachment_data.attachments
+    attachables = attachments.map(&:attachable).compact
+
+    next unless attachables.any?
+    next if attachables.detect { |attachable| !attachable.is_a?(Edition) }
+    next if (attachables.map(&:state) - %w[superseded]).any?
+
+    next if attachment_data.deleted?
+
+    document_id = attachables.last.document.id
+
+    attachment_data.assets.map do |asset|
+      begin
+        am_response = GdsApi.asset_manager.asset(asset.asset_manager_id).to_h
+      rescue GdsApi::HTTPNotFound
+        next
+      end
+
+      variant = asset.variant
+
+      if attachment_data.replaced_by_id
+        if am_response["replacement_id"]
+          replacement = get_rep(am_response["replacement_id"])
+
+          if replacement["draft"]
+            puts "draft_rep - ad: #{attachment_data.id}, d_id: #{document_id}, am_id: #{asset.asset_manager_id}, v: #{variant}, deleted: #{am_response['deleted']}, draft: #{am_response['draft']}, redirect: #{!am_response['redirect_url'].nil?}, #{attachment_data.created_at.year}, rep_del: #{replacement['deleted']}, rep_rep: #{!replacement['replacement_id'].nil?}"
+            file11 << "draft_rep - ad: #{attachment_data.id}, d_id: #{document_id}, am_id: #{asset.asset_manager_id}, v: #{variant}, deleted: #{am_response['deleted']}, draft: #{am_response['draft']}, redirect: #{!am_response['redirect_url'].nil?}, #{attachment_data.created_at.year}, rep_del: #{replacement['deleted']}, rep_rep: #{!replacement['replacement_id'].nil?}" << "\n"
+          else
+            puts "all OK - ad: #{attachment_data.id}, d_id: #{document_id}, am_id: #{asset.asset_manager_id}, deleted: #{am_response['deleted']}, draft: #{am_response['draft']}, redirect: #{!am_response['redirect_url'].nil?}, #{attachment_data.created_at.year}"
+            file1 << "all OK - ad: #{attachment_data.id}, d_id: #{document_id}, am_id: #{asset.asset_manager_id}, deleted: #{am_response['deleted']}, draft: #{am_response['draft']}, redirect: #{!am_response['redirect_url'].nil?}, #{attachment_data.created_at.year}" << "\n"
+          end
+        else
+          puts "missing rep am - ad: #{attachment_data.id}, d_id: #{document_id}, am_id: #{asset.asset_manager_id}, v: #{variant}, deleted: #{am_response['deleted']}, draft: #{am_response['draft']}, redirect: #{!am_response['redirect_url'].nil?}, #{attachment_data.created_at.year}"
+          file2 << "missing rep am - ad: #{attachment_data.id}, d_id: #{document_id}, am_id: #{asset.asset_manager_id}, v: #{variant}, deleted: #{am_response['deleted']}, draft: #{am_response['draft']}, redirect: #{!am_response['redirect_url'].nil?}, #{attachment_data.created_at.year}" << "\n"
+        end
+      else
+        puts "nothing - ad: #{attachment_data.id}, d_id: #{document_id}, am_id: #{asset.asset_manager_id}, v: #{variant}, deleted: #{am_response['deleted']}, draft: #{am_response['draft']}, redirect: #{!am_response['redirect_url'].nil?}, #{attachment_data.created_at.year}, AL: #{am_response['access_limited'].present?}, ALO: #{am_response['access_limited_organisation_ids'].present?}"
+        file3 << "nothing - ad: #{attachment_data.id}, d_id: #{document_id}, am_id: #{asset.asset_manager_id}, v: #{variant}, deleted: #{am_response['deleted']}, draft: #{am_response['draft']}, redirect: #{!am_response['redirect_url'].nil?}, #{attachment_data.created_at.year}, AL: #{am_response['access_limited'].present?}, ALO: #{am_response['access_limited_organisation_ids'].present?}" << "\n"
+      end
+    end
+  end
+
+  file1.close
+  file11.close
+  file2.close
+  file3.close
+end
+
+def get_rep(replacement_id)
+  begin
+    am_response = GdsApi.asset_manager.asset(replacement_id).to_h
+  rescue GdsApi::HTTPNotFound
+    puts "#{attachment_data.id}, #{asset.asset_manager_id}, no replacement found for given ID"
+
+    nil
+  end
+
+  am_response
+end

lib/tasks/attachments/find_attachments_unpublished_in_whitehall_with_missing_redirects_in_asset_manager.rake

@@ -0,0 +1,26 @@
+desc "Identify all AttachmentData instances where attachable is unpublished but there are missing redirects in AM"
+task find_attachments_unpublished_in_whitehall_with_missing_redirects_in_asset_manager: :environment do
+  file = File.open("./lib/tasks/attachments/attachments_without_redirects_in_am.txt", "a")
+
+  CSV.foreach("./lib/tasks/attachments/attachments_on_unpublished_editions.csv", headers: true) do |row|
+    asset_manager_id = row["asset_manager_id"]
+    attachment_data_id = row["attachment_data_id"]
+
+    begin
+      am_response = GdsApi.asset_manager.asset(asset_manager_id).to_h
+    rescue GdsApi::HTTPNotFound
+      next
+    end
+
+    if am_response["redirect_url"].nil?
+      attachment_data = AttachmentData.find(attachment_data_id)
+      document_id = attachment_data&.attachments&.first&.attachable&.document_id
+      variant = attachment_data.assets.select { |a| a["asset_manager_id"] == asset_manager_id }.first.variant
+
+      puts "ad: #{attachment_data_id}, am_id: #{asset_manager_id}, d_id: #{document_id}, var: #{variant}, draft: #{am_response['draft']}, deleted: #{am_response['deleted']}, replaced: #{!am_response['replacement_id'].nil?}, year: #{attachment_data.created_at.year}"
+      file << "#{attachment_data_id},#{asset_manager_id},#{document_id},#{variant},#{am_response['draft']},#{am_response['deleted']},#{!am_response['replacement_id'].nil?},#{attachment_data.created_at.year}" << "\n"
+    end
+  end
+
+  file.close
+end