iiif_print 1.0.0

data/lib/iiif_print.rb

@@ -0,0 +1,136 @@
+require "iiif_print/engine"
+require "iiif_print/errors"
+require "iiif_print/jp2_image_metadata"
+require "iiif_print/image_tool"
+require "iiif_print/text_extraction"
+require "iiif_print/data"
+require "iiif_print/configuration"
+require "iiif_print/base_derivative_service"
+require "iiif_print/jp2_derivative_service"
+require "iiif_print/pdf_derivative_service"
+require "iiif_print/text_extraction_derivative_service"
+require "iiif_print/text_formats_from_alto_service"
+require "iiif_print/tiff_derivative_service"
+require "iiif_print/lineage_service"
+require "iiif_print/metadata"
+require "iiif_print/works_controller_behavior"
+require "iiif_print/jobs/application_job"
+require "iiif_print/blacklight_iiif_search/annotation_decorator"
+require "iiif_print/jobs/child_works_from_pdf_job"
+require "iiif_print/jobs/create_relationships_job"
+require "iiif_print/split_pdfs/pages_into_images_service"
+require "iiif_print/split_pdfs/child_work_creation_from_pdf_service"
+
+module IiifPrint
+  extend ActiveSupport::Autoload
+  autoload :Configuration
+  autoload :CatalogSearchBuilder
+
+  ##
+  # @api public
+  # Exposes the IiifPrint configuration.
+  #
+  # @yield [IiifPrint::Configuration] if a block is passed
+  # @return [IiifPrint::Configuration]
+  # @see IiifPrint::Configuration for configuration options
+  def self.config(&block)
+    @config ||= IiifPrint::Configuration.new
+    yield @config if block
+    @config
+  end
+
+  DEFAULT_MODEL_CONFIGURATION = {
+    # Split a PDF into individual page images and create a new child work for each image.
+    pdf_splitter_job: IiifPrint::Jobs::ChildWorksFromPdfJob,
+    pdf_splitter_service: IiifPrint::SplitPdfs::PagesIntoImagesService,
+    derivative_service_plugins: [
+      IiifPrint::JP2DerivativeService,
+      IiifPrint::PDFDerivativeService,
+      IiifPrint::TextExtractionDerivativeService,
+      IiifPrint::TIFFDerivativeService
+    ]
+  }.freeze
+
+  # This is the record level configuration for PDF split handling.
+  ModelConfig = Struct.new(:pdf_split_child_model, *DEFAULT_MODEL_CONFIGURATION.keys, keyword_init: true)
+
+  # This method is responsible for assisting in the configuration of a "model".
+  #
+  # @example
+  #   class Book < ActiveFedora::Base
+  #     include IiifPrint.model_configuration(
+  #       pdf_split_child_model: Page,
+  #       derivative_service_plugins: [
+  #         IiifPrint::TIFFDerivativeService
+  #       ]
+  #     )
+  #   end
+  #
+  # @param kwargs [Hash<Symbol,Object>] the configuration values that overrides the
+  #        DEFAULT_MODEL_CONFIGURATION.
+  #
+  # @return [Module]
+  #
+  # @see IiifPrint::DEFAULT_MODEL_CONFIGURATION
+  # @todo Because not every job will split PDFs and write to a child model. May want to introduce
+  #       an alternative splitting method to create new filesets on the existing work instead of new child works.
+  def self.model_configuration(**kwargs)
+    Module.new do
+      def iiif_print_config?
+        true
+      end
+
+      # We don't know what you may want in your configuration, but from this gems implementation,
+      # we're going to provide the defaults to ensure that it works.
+      DEFAULT_MODEL_CONFIGURATION.each_pair do |key, default_value|
+        kwargs[key] ||= default_value
+      end
+
+      define_method(:iiif_print_config) do
+        @iiif_print_config ||= ModelConfig.new(**kwargs)
+      end
+    end
+  end
+
+  # @api public
+  #
+  # Map the given work's metadata to the given IIIF version spec's metadata structure.  This
+  # is intended to be a drop-in replacement for `Hyrax::IiifManifestPresenter#manifest_metadata`.
+  #
+  # @param work [Object]
+  # @param version [Integer]
+  # @param fields [Array<IiifPrint::Metadata::Field>, Array<#name, #label>]
+  # @return [Array<Hash>]
+  #
+  # @see specs for expected output
+  #
+  # @see Hyrax::IiifManifestPresenter#manifest_metadata
+  def self.manifest_metadata_for(work:,
+                                 version: config.default_iiif_manifest_version,
+                                 fields: default_fields_for(work),
+                                 current_ability:,
+                                 base_url:)
+    Metadata.build_metadata_for(work: work,
+                                version: version,
+                                fields: fields,
+                                current_ability: current_ability,
+                                base_url: base_url)
+  end
+
+  # Hash is an arbitrary attribute key/value pairs
+  # Struct is a defined set of attribute "keys".  When we favor defined values,
+  # then we are naming the concept and defining the range of potential values.
+  Field = Struct.new(:name, :label, :options, keyword_init: true)
+
+  # @api private
+  # @todo Figure out a way to use a custom label, right now it takes it get rendered from the title.
+  def self.default_fields_for(_work, fields: config.metadata_fields)
+    fields.map do |field|
+      Field.new(
+        name: field.first,
+        label: Hyrax::Renderers::AttributeRenderer.new(field, nil).label,
+        options: field.last
+      )
+    end
+  end
+end

data/lib/tasks/set_child_works.rake

@@ -0,0 +1,22 @@
+# frozen_string_literal: true
+
+namespace :iiif_print do
+  namespace :update do
+    desc 'Make sure all child work models have the correct attribute'
+    task is_child_attribute: [:environment] do
+      Account.find_each do |account|
+        switch!(account.cname) if defined? Account
+        puts "********************** switched to #{account.cname} **********************"
+        Hyrax.config.curation_concerns.each do |cc|
+          puts "********************** checking #{cc}s **********************"
+          next if cc.count.zero?
+
+          cc.find_each(&:save)
+        end
+      rescue StandardError
+        puts "********************** failed to update account #{account.cname} **********************"
+        next
+      end
+    end
+  end
+end

data/spec/.keep.txt

@@ -0,0 +1 @@
+spec dir for RSpec

data/spec/factories/ability.rb

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

data/spec/factories/newspaper_issue.rb

@@ -0,0 +1,7 @@
+# will infer, create a NewspaperIssue object
+FactoryBot.define do
+  factory :newspaper_issue do
+    title { ['Here and There'] }
+    depositor { User.batch_user.user_key }
+  end
+end

data/spec/factories/newspaper_page.rb

@@ -0,0 +1,7 @@
+# will infer, create a NewspaperPage object
+FactoryBot.define do
+  factory :newspaper_page do
+    title { ['Here and There'] }
+    depositor { User.batch_user.user_key }
+  end
+end

data/spec/factories/newspaper_page_solr_document.rb

@@ -0,0 +1,12 @@
+FactoryBot.define do
+  factory :newspaper_page_solr_document, class: SolrDocument do
+    initialize_with do
+      new(id: '123456',
+          title_tesim: ['Page 1'],
+          has_model_ssim: ['NewspaperPage'],
+          issue_id_ssi: 'abc123',
+          file_set_ids_ssim: ['7891011'],
+          thumbnail_path_ss: '/downloads/123456?file=thumbnail')
+    end
+  end
+end

data/spec/factories/newspaper_title.rb

@@ -0,0 +1,8 @@
+# will infer, create a NewspaperTitle object
+FactoryBot.define do
+  factory :newspaper_title do
+    title { ['ACME Press'] }
+    lccn { 'sn2036999999' }
+    depositor { User.batch_user.user_key }
+  end
+end

data/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(IiifPrint::GEM_PATH, 'spec', 'fixtures', 'files')
+      pdf_path = File.join(base, 'minimal-2-page.pdf')
+      new(file: File.open(pdf_path), user: create(:user))
+    end
+  end
+end

data/spec/factories/uploaded_txt_file.rb

@@ -0,0 +1,9 @@
+FactoryBot.define do
+  factory :uploaded_txt_file, class: Hyrax::UploadedFile do
+    initialize_with do
+      base = File.join(IiifPrint::GEM_PATH, 'spec', 'fixtures', 'files')
+      file_path = File.join(base, 'ndnp-sample1-txt.txt')
+      new(file: File.open(file_path), user: create(:user))
+    end
+  end
+end

data/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

data/spec/fixtures/files/README.md

@@ -0,0 +1,7 @@
+## Sample / fixture file manifest
+
+* sample-4page-issue.pdf
+  - original: chicopee-weekly-journal_1856-05-24-p1.pdf
+* page1.tiff
+  - first page from sample-4page-issue.pdf, converted to CCITT Group 4
+    encoded (lossless compressed) TIFF via Ghostscript.