PDF ingest upload Actor/Job implementation

Fixes #33.

* Independent method to create child NewspaperPage works with TIFF for
NewspaperIssueIngest, purpose of which is to be called by job triggered
by Hyrax upload, which is already handling attachment of PDF to issue.

* NewspaperIssueIngest only saves once and without validation for all
child pages added.

* CreateIssuePagesJob and NewspaperWorksUploadActor that calls it,
injected into actor stack before OOTB Hyrax upload actor.

* Admin set of NewspaperIssue work, or default admin set, saved to created
child NewspaperPage objects on PDF upload.

* Factories for upload (pdf), ability, user.

* Add faraday as a testing/development dependency, useful for checking
Fedora in RSpec tests.

* Use :async not :sidekiq as presumed default in tests when disabling
inline after temporary use.

* get Travis-CI to install FITS

.travis.yml

@@ -20,6 +20,10 @@ before_install:
   - gem update --system
   - gem install bundler
   - google-chrome-stable --headless --disable-gpu --no-sandbox --remote-debugging-port=9222 http://localhost &
+  - sudo wget -P /opt/install https://brussels.lib.utah.edu/FITS/fits-1.3.0.zip
+  - sudo unzip /opt/install/fits-1.3.0.zip -d /opt/install/fits-1.3.0
+  - sudo chmod +x /opt/install/fits-1.3.0/*.sh
+  - sudo ln -s /opt/install/fits-1.3.0/fits.sh /usr/local/bin/fits.sh
 
 rvm:
   - 2.5.0

Gemfile

@@ -23,7 +23,9 @@ else
       gem 'rails', github: 'rails/rails'
       ENV['ENGINE_CART_RAILS_OPTIONS'] = '--edge --skip-turbolinks'
     else
+      # rubocop:disable Bundler/DuplicatedGem
       gem 'rails', ENV['RAILS_VERSION']
+      # rubocop:enable Bundler/DuplicatedGem
     end
   end
 
@@ -33,7 +35,9 @@ else
     gem 'responders', '~> 2.0'
     gem 'sass-rails', '>= 5.0'
   when /^4.[01]/
+    # rubocop:disable Bundler/DuplicatedGem
     gem 'sass-rails', '< 5.0'
+    # rubocop:enable Bundler/DuplicatedGem
   end
 end
 # END ENGINE_CART BLOCK

app/actors/newspaper_works/actors/newspaper_works_upload_actor.rb

@@ -0,0 +1,57 @@
+module NewspaperWorks
+  module Actors
+    class NewspaperWorksUploadActor < Hyrax::Actors::BaseActor
+      def create(env)
+        # If NewspaperIssue, we might have a PDF to split...
+        handle_issue_upload(env) if env.curation_concern.class == NewspaperIssue
+        # pass to next actor
+        next_actor.create(env)
+      end
+
+      def update(env)
+        handle_issue_upload(env) if env.curation_concern.class == NewspaperIssue
+        # pass to next actor
+        next_actor.update(env)
+      end
+
+      def default_admin_set
+        AdminSet.find_or_create_default_admin_set_id
+      end
+
+      def queue_job(work, paths, user, admin_set_id)
+        NewspaperWorks::CreateIssuePagesJob.perform_later(
+          work,
+          paths,
+          user,
+          admin_set_id
+        )
+      end
+
+      def handle_issue_upload(env)
+        return unless env.attributes.keys.include? 'uploaded_files'
+        upload_ids = filter_file_ids(env.attributes['uploaded_files'])
+        return if upload_ids.empty?
+        uploads = Hyrax::UploadedFile.find(upload_ids)
+        paths = uploads.map(&method(:upload_path))
+        paths = paths.select { |path| path.end_with?('.pdf') }
+        return if paths.empty?
+        work = env.curation_concern
+        # must persist work to serialize job using it
+        work.save!(validate: false)
+        user = env.current_ability.current_user.user_key
+        env.attributes[:admin_set_id] ||= default_admin_set
+        queue_job(work, paths, user, env.attributes[:admin_set_id])
+      end
+
+      # Given Hyrax::Upload object, return path to file on local filesystem
+      def upload_path(upload)
+        # so many layers to this onion:
+        upload.file.file.file
+      end
+
+      def filter_file_ids(input)
+        Array.wrap(input).select(&:present?)
+      end
+    end
+  end
+end

app/jobs/newspaper_works/create_issue_pages_job.rb

@@ -0,0 +1,17 @@
+module NewspaperWorks
+  # Create child page works for issue
+  class CreateIssuePagesJob < NewspaperWorks::ApplicationJob
+    def perform(work, pdf_paths, user, admin_set_id)
+      # we will need depositor set on work, if it is nil
+      work.depositor ||= user
+      # if we do not have admin_set_id yet, set it on the issue work:
+      work.admin_set_id ||= admin_set_id
+      # create child pages for each page within each PDF uploaded:
+      pdf_paths.each do |path|
+        adapter = NewspaperWorks::Ingest::NewspaperIssueIngest.new(work)
+        adapter.load(path)
+        adapter.create_child_pages
+      end
+    end
+  end
+end

lib/newspaper_works/engine.rb

@@ -8,5 +8,11 @@ module NewspaperWorks
   # Engine Class
   class Engine < ::Rails::Engine
     isolate_namespace NewspaperWorks
+
+    config.to_prepare do
+      # Register actor to handle any NewspaperWorks upload behaviors before
+      #   CreateWithFilesActor gets to them:
+      Hyrax::CurationConcern.actor_factory.insert_before Hyrax::Actors::CreateWithFilesActor, NewspaperWorks::Actors::NewspaperWorksUploadActor
+    end
   end
 end

lib/newspaper_works/ingest/newspaper_issue_ingest.rb

@@ -5,20 +5,31 @@ def import
         # first, handle the PDF itself on the issue...
         super
         # ...then create child works from split pages
-        pages = NewspaperWorks::Ingest::PdfPages.new(path)
+        create_child_pages
+      end
+
+      # Creates child pages with attached TIFF masters, can be called by
+      #   `import`, or independently if `load` is called first.  The
+      #   latter is appropriate if framework is already handling the
+      #   NewspaperIssue file attachment (e.g. Hyrax upload via browser).
+      def create_child_pages
+        pages = NewspaperWorks::Ingest::PdfPages.new(path).to_a
         pages.each_with_index do |tiffpath, idx|
-          new_child_page_with_file(tiffpath, idx)
+          page = new_child_page_with_file(tiffpath, idx)
+          @work.members.push(page)
         end
+        @work.save!(validate: false) unless pages.empty?
       end
 
       def new_child_page_with_file(tiffpath, idx)
         page = NewspaperPage.new
         page.title = [format("Page %<pagenum>i", pagenum: idx + 1)]
+        # Set depositor and admin-set id:
         page.depositor = @work.depositor
-        page.save!
-        @work.members.push(page)
-        @work.save!
+        page.admin_set_id = @work.admin_set_id
         NewspaperPageIngest.new(page).ingest(tiffpath)
+        page.save!
+        page
       end
     end
   end

newspaper_works.gemspec

@@ -28,12 +28,12 @@ SUMMARY
   spec.add_development_dependency 'bixby'
   spec.add_development_dependency 'engine_cart', '~> 2.0'
   spec.add_development_dependency "factory_bot", '~> 4.4'
+  spec.add_development_dependency "faraday"
   spec.add_development_dependency 'fcrepo_wrapper', '~> 0.1'
   spec.add_development_dependency 'rspec'
   spec.add_development_dependency 'rspec-rails', '~> 3.1'
   spec.add_development_dependency 'solr_wrapper', '~> 0.4'
   spec.add_development_dependency 'sqlite3'
 
   spec.add_dependency 'simple_form', '~> 3.2', '<= 3.5.0'
-
 end

spec/actors/newspaper_works/actors/newspaper_works_upload_actor_spec.rb

@@ -0,0 +1,52 @@
+require 'faraday'
+require 'spec_helper'
+
+RSpec.describe NewspaperWorks::Actors::NewspaperWorksUploadActor do
+  let(:issue) { build(:newspaper_issue) }
+  let(:ability) { build(:ability) }
+  let(:uploaded_pdf_file) { create(:uploaded_pdf_file) }
+  let(:uploaded_file_ids) { [uploaded_pdf_file.id] }
+  let(:attributes) { { uploaded_files: uploaded_file_ids } }
+  let(:terminator) { Hyrax::Actors::Terminator.new }
+  let(:env) { Hyrax::Actors::Environment.new(issue, ability, attributes) }
+  let(:middleware) do
+    stack = ActionDispatch::MiddlewareStack.new.tap do |middleware|
+      middleware.use described_class
+    end
+    stack.build(terminator)
+  end
+
+  let(:uploaded_issue) do
+    Rails.application.config.active_job.queue_adapter = :inline
+    middleware.public_send(:create, env)
+    Rails.application.config.active_job.queue_adapter = :async
+    # return work, reloaded, because env.curation_concern will be stale after
+    #   running actor.
+    NewspaperIssue.find(env.curation_concern.id)
+  end
+
+  describe "NewspaperIssue upload of PDF" do
+    # we over-burden one example, because sadly RSpec does not do well with
+    #   shared state across examples (without use of `before(:all)` which is
+    #   mutually exclusive with `let` in practice, and ruffles rubocop's
+    #   overzealous sense of moral duty, speaking of which:
+    # rubocop:disable RSpec/ExampleLength
+    it "correctly creates child pages for issue" do
+      pages = uploaded_issue.members.select { |w| w.class == NewspaperPage }
+      expect(pages.size).to eq 2
+      pages.each_with_index do |page|
+        # Page needs correct admin set:
+        expect(page.admin_set_id).to eq 'admin_set/default'
+        file_sets = page.members.select { |v| v.class == FileSet }
+        expect(file_sets.size).to eq 1
+        files = file_sets[0].files
+        url = files[0].uri.to_s
+        # fetch the thing from Fedora Commons:
+        response = Faraday.get(url)
+        stored_size = response.body.length
+        expect(stored_size).to be > 0
+      end
+    end
+    # rubocop:enable RSpec/ExampleLength
+  end
+end

spec/factories/ability.rb

@@ -0,0 +1,6 @@
+FactoryBot.define do
+  factory :ability do
+    user
+    initialize_with { new(user) }
+  end
+end

spec/factories/uploaded_pdf_file.rb

@@ -0,0 +1,9 @@
+FactoryBot.define do
+  factory :uploaded_pdf_file, class: Hyrax::UploadedFile do
+    initialize_with do
+      base = File.join(NewspaperWorks::GEM_PATH, 'spec', 'fixtures', 'files')
+      pdf_path = File.join(base, 'sample-color-newsletter.pdf')
+      new(file: File.open(pdf_path), user: create(:user))
+    end
+  end
+end

spec/factories/user.rb

@@ -0,0 +1,13 @@
+FactoryBot.define do
+  factory :user do
+    id "skroob"
+    email "spaceballs@example.com"
+    password "password_is_12345"
+    initialize_with do
+      User.find_or_create_by(id: id) do |user|
+        user.email = email
+        user.password = password
+      end
+    end
+  end
+end

spec/lib/newspaper_works/ingest/newspaper_page_ingest_spec.rb

@@ -16,7 +16,7 @@
       adapter = build(:newspaper_page_ingest)
       # Rails.application.config.active_job.queue_adapter = :inline
       adapter.ingest(path)
-      # Rails.application.config.active_job.queue_adapter = :sidekiq
+      # Rails.application.config.active_job.queue_adapter = :async
       file_sets = adapter.work.members.select { |w| w.class == FileSet }
       expect(file_sets[0].title).to contain_exactly 'page1.tiff'
       expect(file_sets.size).to eq 1

spec/spec_helper.rb

@@ -7,15 +7,11 @@
 EngineCart.load_application!
 
 RSpec.configure do |config|
-
   # enable FactoryBot:
   require 'factory_bot'
   config.include FactoryBot::Syntax::Methods
-  # require to load specific factories:
-  require 'factories/newspaper_issue'
-  require 'factories/newspaper_issue_ingest'
-  require 'factories/newspaper_page'
-  require 'factories/newspaper_page_ingest'
+  # auto-detect and load all factories in spec/factories:
+  FactoryBot.find_definitions
 
   # require shared examples
   require 'lib/newspaper_works/ingest/ingest_shared'

test/lib/generators/newspaper_works/install_generator_test.rb

@@ -4,7 +4,7 @@
 module NewspaperWorks
   class InstallGeneratorTest < Rails::Generators::TestCase
     tests InstallGenerator
-    destination Rails.root.join('tmp/generators')
+    destination Rails.root.join('tmp', 'generators')
     setup :prepare_destination
 
     # test "generator runs without errors" do