Add PDF Generator.

.rubocop.yml

@@ -23,6 +23,9 @@ Style/PredicateName:
 Rails/OutputSafety:
   Exclude:
     - 'app/decorators/**/*'
+Metrics/AbcSize:
+  Exclude:
+    - 'app/services/pdf_generator/cover_page_generator.rb'
 Metrics/BlockLength:
   Exclude:
     - 'valhalla/app/models/concerns/valhalla/ability.rb'
@@ -36,13 +39,15 @@ Metrics/BlockLength:
     - 'db/schema.rb'
     - 'lib/tasks/dev.rake'
     - 'app/models/plum_schema.rb'
+    - 'app/services/pdf_generator/cover_page_generator.rb'
 Metrics/MethodLength:
   Exclude:
     - 'db/migrate/**/*'
     - 'app/models/schema/dublin_core.rb'
     - 'app/models/schema/marc_relators.rb'
     - 'app/models/schema/plum/local.rb'
     - 'spec/services/manifest_builder_spec.rb'
+    - 'app/services/pdf_generator/cover_page_generator.rb'
 Metrics/ModuleLength:
   Exclude:
     - 'app/models/schema/dublin_core.rb'

Gemfile

@@ -34,6 +34,7 @@ group :development, :test do
   gem 'coveralls'
   gem "dotenv-rails"
   gem "factory_girl_rails"
+  gem 'pdf-reader', github: 'yob/pdf-reader'
   gem "pry-byebug"
   gem "pry-rails"
   gem 'rails-controller-testing'
@@ -77,3 +78,4 @@ gem 'valhalla', path: 'valhalla'
 source 'https://rails-assets.org' do
   gem 'rails-assets-bootstrap-select', '1.12.2'
 end
+gem 'prawn'

Gemfile.lock

@@ -24,6 +24,17 @@ GIT
       activesupport (>= 4)
       iiif-presentation (~> 0.2.0)
 
+GIT
+  remote: git://github.com/yob/pdf-reader.git
+  revision: bb7747a3cb4279ecd963185f357dbd718f9d20f3
+  specs:
+    pdf-reader (2.0.0)
+      Ascii85 (~> 1.0.0)
+      afm (~> 0.2.1)
+      hashery (~> 2.0)
+      ruby-rc4
+      ttfunk
+
 GIT
   remote: https://github.com/cbeer/devise-guests.git
   revision: 0999e0bf31841640ad52331e8db14c3f0e1c9765
@@ -66,6 +77,7 @@ GEM
   remote: https://rubygems.org/
   remote: https://rails-assets.org/
   specs:
+    Ascii85 (1.0.2)
     aasm (4.12.2)
       concurrent-ruby (~> 1.0)
     actioncable (5.1.1)
@@ -126,6 +138,7 @@ GEM
       tzinfo (~> 1.1)
     addressable (2.5.1)
       public_suffix (~> 2.0, >= 2.0.2)
+    afm (0.2.2)
     almond-rails (0.1.0)
       rails (>= 4.2, < 6)
     arel (8.0.0)
@@ -321,6 +334,7 @@ GEM
     hamster (3.0.0)
       concurrent-ruby (~> 1.0)
     hashdiff (0.3.4)
+    hashery (2.1.2)
     hashie (3.5.6)
     honeybadger (3.1.2)
     httmultiparty (0.3.16)
@@ -457,12 +471,16 @@ GEM
     parallel (1.12.0)
     parser (2.4.0.0)
       ast (~> 2.2)
+    pdf-core (0.7.0)
     pg (0.21.0)
     poltergeist (1.15.0)
       capybara (~> 2.1)
       cliver (~> 0.3.1)
       websocket-driver (>= 0.2.0)
     powerpack (0.1.1)
+    prawn (2.2.2)
+      pdf-core (~> 0.7.0)
+      ttfunk (~> 1.5)
     pry (0.10.4)
       coderay (~> 1.1.0)
       method_source (~> 0.8.1)
@@ -582,6 +600,7 @@ GEM
       multipart-post
       oauth2
     ruby-progressbar (1.8.1)
+    ruby-rc4 (0.1.5)
     ruby_dep (1.5.0)
     ruby_tika_app (1.5.0)
       open4
@@ -660,6 +679,7 @@ GEM
     title (0.0.7)
       i18n
       rails (>= 3.1)
+    ttfunk (1.5.0)
     twitter-typeahead-rails (0.11.1.pre.corejavascript)
       actionpack (>= 3.1)
       jquery-rails
@@ -725,8 +745,10 @@ DEPENDENCIES
   normalize-rails
   omniauth-cas
   openseadragon
+  pdf-reader!
   pg
   poltergeist
+  prawn
   pry-byebug
   pry-rails
   pul_metadata_services!

app/change_sets/scanned_resource_change_set.rb

@@ -21,6 +21,7 @@ class ScannedResourceChangeSet < Valkyrie::ChangeSet
   property :state, multiple: false, required: true, default: BookWorkflow.aasm.initial_state.to_s
   property :read_groups, multiple: true, required: false
   property :workflow_note, multiple: true, required: false, default: []
+  property :file_metadata, multiple: true, required: false, default: []
   # Virtual Attributes
   property :refresh_remote_metadata, virtual: true, multiple: false
   property :files, virtual: true, multiple: true, required: false

app/controllers/scanned_resources_controller.rb

@@ -39,6 +39,18 @@ def manifest
     end
   end
 
+  def pdf
+    change_set = change_set_class.new(find_resource(params[:id])).prepopulate!
+    authorize! :pdf, change_set.resource
+    pdf_file = PDFGenerator.new(resource: change_set.resource, storage_adapter: Valkyrie::StorageAdapter.find(:derivatives)).render
+    change_set_persister.buffer_into_index do |buffered_changeset_persister|
+      change_set.validate(file_metadata: [pdf_file])
+      change_set.sync
+      buffered_changeset_persister.save(change_set: change_set)
+    end
+    redirect_to valhalla.download_path(resource_id: change_set.id, id: pdf_file.id)
+  end
+
   def selected_file_params
     params[:selected_files].to_unsafe_h
   end

app/models/file_set.rb

@@ -3,7 +3,6 @@ class FileSet < Valhalla::Resource
   include Valkyrie::Resource::AccessControls
   attribute :id, Valkyrie::Types::ID.optional
   attribute :title, Valkyrie::Types::Set
-  attribute :member_ids, Valkyrie::Types::Array
   attribute :file_metadata, Valkyrie::Types::Set.member(FileMetadata.optional)
   attribute :depositor
 

app/models/scanned_resource.rb

@@ -11,6 +11,7 @@ class ScannedResource < Valhalla::Resource
   attribute :logical_structure, Valkyrie::Types::Array.member(Structure.optional).optional
   attribute :pending_uploads, Valkyrie::Types::Array.member(PendingUpload)
   attribute :workflow_note, Valkyrie::Types::Array.member(WorkflowNote).optional
+  attribute :file_metadata, Valkyrie::Types::Set.member(FileMetadata.optional)
 
   def to_s
     "#{human_readable_type}: #{title.to_sentence}"
@@ -20,6 +21,12 @@ def primary_imported_metadata
     Array.wrap(imported_metadata).first || ImportedMetadata.new
   end
 
+  def pdf_file
+    file_metadata.find do |file|
+      file.mime_type == ["application/pdf"]
+    end
+  end
+
   def title
     primary_imported_metadata.title.present? ? primary_imported_metadata.title : @title
   end

app/services/file_appender.rb

@@ -13,7 +13,7 @@ def append_to(resource)
     return updated_files unless updated_files.empty?
 
     file_sets = build_file_sets || file_nodes
-    if resource.respond_to?(:file_metadata)
+    if file_set?(resource)
       resource.file_metadata += file_sets
     else
       resource.member_ids += file_sets.map(&:id)
@@ -23,6 +23,10 @@ def append_to(resource)
     file_sets
   end
 
+  def file_set?(resource)
+    resource.respond_to?(:file_metadata) && !resource.respond_to?(:member_ids)
+  end
+
   def update_files(resource, files)
     files.select { |file| file.is_a?(Hash) }.map do |file|
       node = resource.file_metadata.select { |x| x.id.to_s == file.keys.first }.first

app/services/manifest_builder.rb

@@ -30,7 +30,7 @@ def work_presenters
 
     def file_set_presenters
       @file_set_presenters ||= leaf_nodes.map do |node|
-        LeafNode.new(node)
+        LeafNode.new(node, self)
       end
     end
 
@@ -106,10 +106,11 @@ def id
   end
 
   class LeafNode
-    attr_reader :resource
+    attr_reader :resource, :parent_node
     delegate :query_service, to: :metadata_adapter
-    def initialize(resource)
+    def initialize(resource, parent_node)
       @resource = resource
+      @parent_node = parent_node
     end
 
     delegate :id, to: :resource

app/services/pdf_generator.rb

@@ -0,0 +1,58 @@
+# frozen_string_literal: true
+class PDFGenerator
+  attr_reader :resource, :storage_adapter
+  def initialize(resource:, storage_adapter:)
+    @resource = resource
+    @storage_adapter = storage_adapter
+  end
+
+  def render
+    CoverPageGenerator.new(self).apply(prawn_document)
+    canvas_downloaders.each_with_index do |downloader, _index|
+      prawn_document.start_new_page layout: downloader.layout
+      page_size = [Canvas::LETTER_WIDTH, Canvas::LETTER_HEIGHT]
+      page_size.reverse! unless downloader.portrait?
+      prawn_document.image downloader.download, width: downloader.width, height: downloader.height, fit: page_size
+    end
+    prawn_document.render_file(tmp_file.path)
+    build_node
+  end
+
+  def build_node
+    file = IngestableFile.new(file_path: tmp_file.path, mime_type: 'application/pdf', original_filename: 'derivative_pdf.pdf')
+    node = FileMetadata.for(file: file).new(id: SecureRandom.uuid)
+    stored_file = storage_adapter.upload(resource: node, file: file)
+    node.file_identifiers = stored_file.id
+    node
+  end
+
+  def prawn_document
+    @prawn_document ||= Prawn::Document.new(prawn_options)
+  end
+
+  def prawn_options
+    default_options = { margin: 0 }
+    default_options[:page_layout] = :portrait if canvas_downloaders.first
+    default_options
+  end
+
+  def canvas_images
+    @canvas_images ||= manifest['sequences'][0]['canvases'].map { |x| x['images'][0] }.map do |x|
+      Canvas.new(x)
+    end
+  end
+
+  def manifest
+    @manifest ||= ManifestBuilder.new(resource).build
+  end
+
+  def canvas_downloaders
+    @canvas_images ||= canvas_images.map do |image|
+      CanvasDownloader.new(image, quality: resource.pdf_type.first)
+    end
+  end
+
+  def tmp_file
+    @tmp_file ||= Tempfile.new("pdf")
+  end
+end

app/services/pdf_generator/canvas.rb

@@ -0,0 +1,25 @@
+# frozen_string_literal: true
+class PDFGenerator
+  class Canvas
+    # Letter width/height in points for a PDF.
+    LETTER_WIDTH = PDF::Core::PageGeometry::SIZES["LETTER"].first
+    LETTER_HEIGHT = PDF::Core::PageGeometry::SIZES["LETTER"].last
+    BITONAL_SIZE = 2000
+    attr_reader :canvas
+    def initialize(canvas)
+      @canvas = canvas
+    end
+
+    def width
+      canvas["resource"]["width"].to_i
+    end
+
+    def height
+      canvas["resource"]["height"].to_i
+    end
+
+    def url
+      canvas["resource"]["service"]["@id"]
+    end
+  end
+end

app/services/pdf_generator/canvas_downloader.rb

@@ -0,0 +1,63 @@
+# frozen_string_literal: true
+class PDFGenerator
+  class CanvasDownloader
+    attr_reader :canvas
+    delegate :width, :height, to: :canvas
+    def initialize(canvas, quality: "grey")
+      @canvas = canvas
+      @quality = quality
+    end
+
+    def download
+      open(canvas_url, 'rb')
+    end
+
+    def layout
+      if portrait?
+        :portrait
+      else
+        :landscape
+      end
+    end
+
+    def quality
+      if @quality == 'gray'
+        'grey'
+      else
+        'default'
+      end
+    end
+
+    def portrait?
+      canvas.width <= canvas.height
+    end
+
+    private
+
+      def canvas_url
+        "#{canvas.url}/full/#{max_width},#{max_height}/0/#{quality}.#{format}"
+      end
+
+      def format
+        bitonal? ? 'png' : 'jpg'
+      end
+
+      def max_width
+        return Canvas::BITONAL_SIZE if bitonal?
+        [(Canvas::LETTER_WIDTH * scale_factor).round, canvas.width].min
+      end
+
+      def max_height
+        return Canvas::BITONAL_SIZE if bitonal?
+        [(Canvas::LETTER_HEIGHT * scale_factor).round, canvas.height].min
+      end
+
+      def scale_factor
+        1.5
+      end
+
+      def bitonal?
+        quality == 'bitonal'
+      end
+  end
+end