achydra-950: monthly cron to delete stale pending works

config/schedule.rb

@@ -32,3 +32,8 @@
 every :day, at: '12am' do
   rake 'resque:restart_workers', email_subject: 'Resque workers restart'
 end
+
+# Delete stale pending works once a month
+every :month do
+  rake 'ac:delete_stale_pending_works', email_subject: 'Delete stale pending works'
+end

lib/tasks/ac.rake

@@ -44,4 +44,38 @@ namespace :ac do
       puts Rainbow('Incorrect arguments. Pass output=/path/to/file').red
     end
   end
+
+  task delete_stale_pending_works: :environment do
+    log = Rails.logger
+    users = User.all
+    users.each do |current_user|
+      deposits = current_user.deposits.where('created_at < ?', 6.months.ago)
+      identifiers = deposits.map(&:hyacinth_identifier).compact
+      if identifiers.present?
+        results = AcademicCommons.search do |params|
+          identifiers = identifiers.map { |i| "\"#{i}\"" }.join(' OR ')
+                                   .prepend('(').concat(')')
+          params.filter('fedora3_pid_ssi', identifiers)
+          params.aggregators_only
+          params.field_list('fedora3_pid_ssi')
+        end
+
+        hyacinth_ids_in_ac = results.documents.map { |d| d[:fedora3_pid_ssi] }
+      else
+        hyacinth_ids_in_ac = []
+      end
+
+      pending_works = deposits.to_a.keep_if do |deposit|
+        hyacinth_id = deposit.hyacinth_identifier
+        hyacinth_id.blank? || !hyacinth_ids_in_ac.include?(hyacinth_id)
+      end
+
+      log.info "===Deleting #{pending_works.count} stale pending work(s)===" unless pending_works.empty?
+
+      pending_works.each do |pending_work|
+        log.info pending_work.inspect
+        pending_work.destroy
+      end
+    end
+  end
 end

lib/tasks/resque.rake

@@ -102,8 +102,7 @@ namespace :resque do
       count.times do |index|
         number = index + 1
         pidfile = Rails.root.join('tmp/pids/', "resque_work_#{number}.pid").to_s
-        _stdout_str, _stderr_str, status = Open3.capture3("RAILS_ENV=#{rails_env}  QUEUE=\"#{queues}\"  PIDFILE=#{pidfile} BACKGROUND=yes VERBOSE=1 INTERVAL=#{interval} rake resque:work >> #{out} 2>> #{err}")
-        puts "Worker #{number} started, status: #{status}"
+        Open3.capture3("RAILS_ENV=#{rails_env}  QUEUE=\"#{queues}\"  PIDFILE=#{pidfile} BACKGROUND=yes VERBOSE=1 INTERVAL=#{interval} rake resque:work >> #{out} 2>> #{err}")
       end
     end
   end

spec/factories/deposit.rb

@@ -15,12 +15,18 @@
     license { 'https://creativecommons.org/licenses/by/4.0/' }
     previously_published { true }
     current_student { false }
+    created_at { Time.zone.now }
+    hyacinth_identifier {}
 
     after(:build) do |deposit|
       deposit.files.attach(
         io: File.open(Rails.root.join('spec', 'fixtures', 'test_file.txt')),
         filename: 'test_file.txt'
       )
     end
+
+    trait :with_user do
+      association :user, factory: :user
+    end
   end
 end

spec/factories/user.rb

@@ -3,7 +3,7 @@
     uid { 'tu123' }
     first_name { 'Test' }
     last_name { 'User' }
-    email { '[email protected]' }
+    sequence(:email) { |n| "tu123#{n}@columbia.edu" }
   end
 
   factory :admin, class: User do

spec/features/upload_spec.rb

@@ -140,7 +140,7 @@
       it 'sends student reminder email' do
         email = ActionMailer::Base.deliveries.pop
         expect(email.subject).to eql 'Department approval may be needed'
-        expect(email.to).to include '[email protected]'
+        expect(email.to).to include User.first.email
       end
     end
 

spec/tasks/ac/pending_works_spec.rb

@@ -0,0 +1,92 @@
+# frozen_string_literal: true
+
+require 'rails_helper'
+
+describe 'rake ac:delete_stale_pending_works', type: :task do
+  context 'when a deposit with no hyacinth identifier' do
+    let!(:deposit_fresh) { FactoryBot.create(:deposit, :with_user) }
+    let!(:deposit_stale) do
+      FactoryBot.create(:deposit, :with_user, created_at: 7.months.ago)
+    end
+
+    describe 'that is less than 6 months old' do
+      it 'does not delete the deposit' do
+        task.execute
+        expect(Deposit.exists?(deposit_fresh.id)).to be true
+      end
+    end
+
+    describe 'that is over 6 months old' do
+      it 'deletes the deposit' do
+        task.execute
+        expect { deposit_stale.reload }.to raise_error(ActiveRecord::RecordNotFound)
+      end
+    end
+  end
+
+  context 'when a deposit with an identifier that is unindexed' do
+    let(:empty_response) { wrap_solr_response_data('response' => { 'docs' => [] }) }
+    let!(:deposit_fresh) do
+      FactoryBot.create(:deposit, :with_user, hyacinth_identifier: 'unindexed')
+    end
+    let!(:deposit_stale) do
+      FactoryBot.create(:deposit, :with_user, created_at: 7.months.ago, hyacinth_identifier: 'unindexed')
+    end
+
+    before do
+      allow(AcademicCommons).to receive(:search).and_return(empty_response)
+    end
+
+    describe 'and is less than 6 months old' do
+      it 'does not delete the deposit' do
+        task.execute
+        expect(Deposit.exists?(deposit_fresh.id)).to be true
+      end
+    end
+
+    describe 'and is over 6 months old' do
+      it 'deletes the deposit' do
+        task.execute
+        expect { deposit_stale.reload }.to raise_error(ActiveRecord::RecordNotFound)
+      end
+    end
+  end
+
+  context 'when a deposit with an identifier that is indexed' do
+    let(:solr_response) do
+      wrap_solr_response_data(
+        'response' => {
+          'docs' => [
+            { 'id' => '10.7916/TESTDOC10', 'title_ssi' => 'First Test Document', 'object_state_ssi' => 'A',
+              'cul_doi_ssi' => '10.7916/TESTDOC10', 'fedora3_pid_ssi' => 'actest:10', 'genre_ssim' => '',
+              'publisher_doi_ssi' => '', 'free_to_read_start_date_ssi' => Date.current.tomorrow.strftime('%Y-%m-%d') }
+          ]
+        }
+      )
+    end
+    let!(:deposit_fresh) do
+      FactoryBot.create(:deposit, :with_user, hyacinth_identifier: 'actest:10')
+    end
+    let!(:deposit_stale) do
+      FactoryBot.create(:deposit, :with_user, created_at: 7.months.ago, hyacinth_identifier: 'actest:10')
+    end
+
+    before do
+      allow(AcademicCommons).to receive(:search).and_return(solr_response)
+    end
+
+    describe 'and is less than 6 months old' do
+      it 'does not delete the deposit' do
+        task.execute
+        expect(Deposit.exists?(deposit_fresh.id)).to be true
+      end
+    end
+
+    describe 'and is over 6 months old' do
+      it 'does not delete the deposit' do
+        task.execute
+        expect(Deposit.exists?(deposit_stale.id)).to be true
+      end
+    end
+  end
+end