newspaper_works 0.1.0

data/app/jobs/newspaper_works/compose_issue_pdf_job.rb

@@ -0,0 +1,13 @@
+module NewspaperWorks
+  # Compose and attach a multi-page PDF from constituent pages, if ready
+  #   (if not ready, job retry requires Rails >= 5.1)
+  class ComposeIssuePDFJob < NewspaperWorks::ApplicationJob
+    retry_on NewspaperWorks::PagesNotReady,
+             wait: :exponentially_longer,
+             attempts: 8
+
+    def perform(issue)
+      NewspaperWorks::IssuePDFComposer.new(issue).compose
+    end
+  end
+end

data/app/jobs/newspaper_works/create_issue_pages_job.rb

@@ -0,0 +1,19 @@
+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
+      # re-save pages so that parent and sibling relationships are indexed
+      work.pages.each(&:save)
+    end
+  end
+end

data/app/mailers/newspaper_works/application_mailer.rb

@@ -0,0 +1,8 @@
+# Application Mailer
+module NewspaperWorks
+  # Application Mailer Class
+  class ApplicationMailer < ActionMailer::Base
+    default from: 'from@example.com'
+    layout 'mailer'
+  end
+end

data/app/models/concerns/newspaper_works/blacklight_iiif_search/annotation_behavior.rb

@@ -0,0 +1,82 @@
+# customize behavior for IiifSearch
+module NewspaperWorks
+  module BlacklightIiifSearch
+    module AnnotationBehavior
+      ##
+      # Create a URL for the annotation
+      # use a Hyrax-y URL syntax:
+      # protocol://host:port/concern/model_type/work_id/manifest/canvas/file_set_id/annotation/index
+      # @return [String]
+      def annotation_id
+        "#{base_url}/manifest/canvas/#{file_set_id}/annotation/#{hl_index}"
+      end
+
+      ##
+      # Create a URL for the canvas that the annotation refers to
+      # match the Hyrax default canvas URL syntax:
+      # protocol://host:port/concern/model_type/work_id/manifest/canvas/file_set_id
+      # @return [String]
+      def canvas_uri_for_annotation
+        "#{base_url}/manifest/canvas/#{file_set_id}#{coordinates}"
+      end
+
+      private
+
+        ##
+        # return a string like "#xywh=100,100,250,20"
+        # corresponding to coordinates of query term on image
+        # @return [String]
+        def coordinates
+          return default_coords if query.blank?
+          coords_json = fetch_and_parse_coords
+          return default_coords unless coords_json && coords_json['coords']
+          query_terms = query.split(' ').map(&:downcase)
+          matches = coords_json['coords'].select do |k, _v|
+            k.downcase =~ /(#{query_terms.join('|')})/
+          end
+          return default_coords if matches.blank?
+          coords_array = matches.values.flatten(1)[hl_index]
+          return default unless coords_array
+          "#xywh=#{coords_array.join(',')}"
+        end
+
+        ##
+        # return the JSON word-coordinates file contents
+        # @return [JSON]
+        def fetch_and_parse_coords
+          coords = NewspaperWorks::Data::WorkDerivatives.new(file_set_id).data('json')
+          return nil if coords.blank?
+          begin
+            JSON.parse(coords)
+          rescue JSON::ParserError
+            nil
+          end
+        end
+
+        ##
+        # a default set of coordinates
+        # @return [String]
+        def default_coords
+          '#xywh=0,0,0,0'
+        end
+
+        ##
+        # the base URL for the Newspaper object
+        # use polymorphic_url, since we deal with multiple object types
+        # @return [String]
+        def base_url
+          host = controller.request.base_url
+          controller.polymorphic_url(parent_document, host: host, locale: nil)
+        end
+
+        ##
+        # return the first file set id
+        # @return [String]
+        def file_set_id
+          file_set_ids = document['file_set_ids_ssim']
+          raise "#{self.class}: NO FILE SET ID" if file_set_ids.blank?
+          file_set_ids.first
+        end
+    end
+  end
+end

data/app/models/concerns/newspaper_works/blacklight_iiif_search/search_behavior.rb

@@ -0,0 +1,27 @@
+# customize behavior for IiifSearch
+module NewspaperWorks
+  module BlacklightIiifSearch
+    module SearchBehavior
+      ##
+      # params to limit the search to items that are children of item
+      # modified to make search field conditional on parent object class
+      # @return [Hash]
+      def object_relation_solr_params
+        parent_model = parent_document['has_model_ssim'].find do |v|
+          v.include?('Newspaper')
+        end
+        solr_field_for_search = case parent_model
+                                when 'NewspaperPage'
+                                  'id'
+                                when 'NewspaperIssue'
+                                  'issue_id_ssi'
+                                when 'NewspaperArticle'
+                                  'article_ids_ssim'
+                                else
+                                  iiif_config[:object_relation_field]
+                                end
+        { solr_field_for_search => id }
+      end
+    end
+  end
+end

data/app/models/concerns/newspaper_works/newspaper_core_metadata.rb

@@ -0,0 +1,67 @@
+# module comment...
+module NewspaperWorks
+  # core metadata for newspaper models
+  module NewspaperCoreMetadata
+    extend ActiveSupport::Concern
+
+    included do
+      # common metadata for Newspaper title, issue, article; fields
+      # that are not in ::Hyrax::BasicMetadata are enumerated here.
+
+      # - Alternative Title
+      property(
+        :alternative_title,
+        predicate: ::RDF::Vocab::DC.alternative,
+        multiple: true
+      ) do |index|
+        index.as :stored_searchable
+      end
+
+      #  - Place of Publication
+      property(
+        :place_of_publication,
+        predicate: ::RDF::Vocab::MARCRelators.pup,
+        multiple: true,
+        class_name: Hyrax::ControlledVocabularies::Location
+      ) do |index|
+        index.as :stored_searchable
+      end
+
+      # - ISSN
+      property(
+        :issn,
+        predicate: ::RDF::Vocab::Identifiers.issn,
+        multiple: false
+      ) do |index|
+        index.as :stored_searchable, :facetable
+      end
+
+      # - LCCN
+      property(
+        :lccn,
+        predicate: ::RDF::Vocab::Identifiers.lccn,
+        multiple: false
+      ) do |index|
+        index.as :stored_searchable, :facetable
+      end
+
+      # - OCLC Number
+      property(
+        :oclcnum,
+        predicate: ::RDF::Vocab::BIBO.oclcnum,
+        multiple: false
+      ) do |index|
+        index.as :stored_searchable, :facetable
+      end
+
+      # Holding location (held by):
+      property(
+        :held_by,
+        predicate: ::RDF::Vocab::BF2.heldBy,
+        multiple: false
+      ) do |index|
+        index.as :stored_searchable, :facetable
+      end
+    end
+  end
+end

data/app/models/concerns/newspaper_works/place_of_publication_behavior.rb

@@ -0,0 +1,15 @@
+# module comment...
+module NewspaperWorks
+  # setting this allows us to use GeoNames autocomplete for place_of_publication property
+  # needs to be included in models after Hyrax::BasicMetadata
+  module PlaceOfPublicationBehavior
+    extend ActiveSupport::Concern
+
+    included do
+      self.controlled_properties = [:place_of_publication]
+      accepts_nested_attributes_for :place_of_publication,
+                                    reject_if: proc { |attributes| attributes[:id].blank? },
+                                    allow_destroy: true
+    end
+  end
+end

data/app/models/concerns/newspaper_works/scanned_media_metadata.rb

@@ -0,0 +1,43 @@
+# Scanned Media: Shared Metadata
+module NewspaperWorks
+  # scanned media metadata for newspaper models (e.g. page, article images)
+  module ScannedMediaMetadata
+    extend ActiveSupport::Concern
+
+    included do
+      # common descriptive metadata properties for scanned media like pages
+      # that do not already have implementation in Hyrax::BasicMetadata
+
+      # - Label
+      #   (implemented by Hyrax::Metadata as :title, we omit here)
+      # - Text direction
+      property(
+        :text_direction,
+        predicate: ::RDF::Vocab::OA.textDirection,
+        multiple: false
+      ) do |index|
+        index.as :stored_searchable
+      end
+
+      # - Page Number
+      property(
+        :page_number,
+        predicate: ::RDF::Vocab::SCHEMA.pagination,
+        multiple: false
+      ) do |index|
+        index.as :stored_searchable
+      end
+
+      # - Section
+      property(
+        :section,
+        predicate: ::RDF::Vocab::BIBO.section,
+        multiple: false
+      ) do |index|
+        index.as :stored_searchable
+      end
+
+      # - Reel # TBD TODO needs predicate TBD
+    end
+  end
+end

data/app/models/concerns/newspaper_works/solr/document.rb

@@ -0,0 +1,25 @@
+module NewspaperWorks::Solr::Document
+  SOLR_NAMES = %w[alternative_title genre place_of_publication
+                  issn lccn oclcnum held_by text_direction
+                  page_number section author photographer
+                  volume issue_number geographic_coverage
+                  extent publication_date height width
+                  edition_number edition_name frequency preceded_by
+                  succeeded_by publication_date_start
+                  publication_date_end].freeze
+
+  def method_missing(m, *args, &block)
+    super unless SOLR_NAMES.include? m.to_s
+    self[Solrizer.solr_name(m.to_s)]
+  end
+
+  def respond_to_missing?(method_name, include_private = false)
+    SOLR_NAMES.include?(method_name.to_s) || super
+  end
+
+  # TODO: figure out if there is a cleaner way to get this
+  #       adding file_set_ids to SOLR_NAMES does not work
+  def file_set_ids
+    self['file_set_ids_ssim']
+  end
+end

data/app/models/file_set.rb

@@ -0,0 +1,10 @@
+# ---
+# newspaper_works needs a fileset compatible with and mixing in
+# ::Hyrax::FileSetBehavior, since Hyrax does not define such a class.
+# Typically, this is provided by boilerplate Hyrax generates into an
+# app, but this engine provides a compatible FileSet class for use
+# and instantiaion by its models.
+# ---
+class FileSet < ActiveFedora::Base
+  include ::Hyrax::FileSetBehavior
+end

data/app/models/newspaper_article.rb

@@ -0,0 +1,158 @@
+# Newspaper Article Cass
+class NewspaperArticle < ActiveFedora::Base
+  # WorkBehavior mixes in minimal ::Hyrax::CoreMetadata fields of
+  # depositor, title, date_uploaded, and date_modified.
+  # https://samvera.github.io/customize-metadata-model.html#core-metadata
+  include ::Hyrax::WorkBehavior
+  include NewspaperWorks::NewspaperCoreMetadata
+  include NewspaperWorks::ScannedMediaMetadata
+
+  self.indexer = NewspaperArticleIndexer
+
+  # containment/aggregation:
+  self.valid_child_concerns = [NewspaperPage]
+
+  validates_with NewspaperWorks::PublicationDateValidator
+
+  # Validation and required fields:
+  validates :title, presence: {
+    message: 'A newspaper article requires a title.'
+  }
+
+  # TODO: Implement validations
+  # validates :resource_type, presence: {
+  #   message: 'A newspaper article requires a resource type.'
+  # }
+  # validates :language, presence: {
+  #   message: 'A newspaper article requires a language.'
+  # }
+  # validates :held_by, presence: {
+  #   message: 'A newspaper article requires a holding location.'
+  # }
+
+  # == Type-specific properties ==
+
+  # TODO: DRY on the indexing of fields, the index block is repetitive...
+
+  # TODO: Reel #: https://github.com/samvera-labs/uri_selection_wg/issues/2
+
+  # - Genre
+  property(
+    :genre,
+    predicate: ::RDF::Vocab::EDM.hasType,
+    multiple: true
+  ) do |index|
+    index.as :stored_searchable, :facetable
+  end
+
+  # - Author
+  property(
+    :author,
+    predicate: ::RDF::Vocab::MARCRelators.aut,
+    multiple: true
+  ) do |index|
+    index.as :stored_searchable, :facetable
+  end
+
+  # - Photographer
+  property(
+    :photographer,
+    predicate: ::RDF::Vocab::MARCRelators.pht,
+    multiple: true
+  ) do |index|
+    index.as :stored_searchable, :facetable
+  end
+
+  # - Volume
+  property(
+    :volume,
+    predicate: ::RDF::Vocab::BIBO.volume,
+    multiple: false
+  ) do |index|
+    index.as :stored_searchable
+  end
+
+  # - Edition name
+  property(
+    :edition_name,
+    predicate: ::RDF::Vocab::BF2.editionStatement,
+    multiple: false
+  ) do |index|
+    index.as :stored_searchable
+  end
+
+  # - Edition number / enumeration
+  property(
+    :edition_number,
+    predicate: ::RDF::Vocab::BF2.editionEnumeration,
+    multiple: false
+  ) do |index|
+    index.as :stored_searchable
+  end
+
+  # - Issue
+  property(
+    :issue_number,
+    predicate: ::RDF::Vocab::BIBO.issue,
+    multiple: false
+  ) do |index|
+    index.as :stored_searchable
+  end
+
+  # - Geographic coverage
+  property(
+    :geographic_coverage,
+    predicate: ::RDF::Vocab::DC.spatial,
+    multiple: true
+  ) do |index|
+    index.as :stored_searchable, :facetable
+  end
+
+  # - Extent
+  property(
+    :extent,
+    predicate: ::RDF::Vocab::DC.extent,
+    multiple: false
+  ) do |index|
+    index.as :stored_searchable
+  end
+
+  #  - publication date
+  property(
+    :publication_date,
+    predicate: ::RDF::Vocab::DC.issued,
+    multiple: false
+  ) do |index|
+    index.type :date
+    index.as :stored_sortable
+  end
+
+  # TODO: Add Reel number: https://github.com/samvera-labs/uri_selection_wg/issues/2
+
+  # BasicMetadata must be included last
+  include ::Hyrax::BasicMetadata
+
+  # for GeoNames autocomplete lookup
+  include NewspaperWorks::PlaceOfPublicationBehavior
+
+  # relationship methods:
+
+  def pages
+    members.select { |v| v.instance_of?(NewspaperPage) }
+  end
+
+  def issue
+    issues = member_of.select { |v| v.instance_of?(NewspaperIssue) }
+    issues[0] unless issues.empty?
+  end
+
+  def publication
+    issue = self.issue
+    issue.publication unless issue.nil?
+  end
+
+  def container
+    pages = self.pages
+    pages.first.container unless pages.empty?
+  end
+end

data/app/models/newspaper_container.rb

@@ -0,0 +1,86 @@
+# Newspaper Container
+class NewspaperContainer < ActiveFedora::Base
+  # WorkBehavior mixes in minimal ::Hyrax::CoreMetadata fields of
+  # depositor, title, date_uploaded, and date_modified.
+  # https://samvera.github.io/customize-metadata-model.html#core-metadata
+  include ::Hyrax::WorkBehavior
+  include NewspaperWorks::NewspaperCoreMetadata
+
+  self.indexer = NewspaperContainerIndexer
+
+  # containment/aggregation:
+  self.valid_child_concerns = [NewspaperPage]
+
+  # Validation and required fields:
+  validates :title, presence: {
+    message: 'A newspaper container requires a title.'
+  }
+
+  validates_with NewspaperWorks::PublicationDateStartEndValidator
+
+  # TODO: Implement validations
+  # validates :resource_type, presence: {
+  #   message: 'A newspaper article requires a resource type.'
+  # }
+  # validates :language, presence: {
+  #   message: 'A newspaper article requires a language.'
+  # }
+  # validates :held_by, presence: {
+  #   message: 'A newspaper article requires a holding location.'
+  # }
+
+  # == Type-specific properties ==
+
+  # TODO: DRY on the indexing of fields, the index block is repetative...
+
+  #  - Type (TODO: make a behavior mixin for common fields)
+
+  # - Extent
+  property(
+    :extent,
+    predicate: ::RDF::Vocab::DC.extent,
+    multiple: false
+  ) do |index|
+    index.as :stored_searchable
+  end
+
+  #  - publication date start
+  property(
+    :publication_date_start,
+    predicate: ::RDF::Vocab::SCHEMA.startDate,
+    multiple: false
+  ) do |index|
+    index.type :date
+    index.as :stored_sortable
+  end
+
+  #  - publication date end
+  property(
+    :publication_date_end,
+    predicate: ::RDF::Vocab::SCHEMA.endDate,
+    multiple: false
+  ) do |index|
+    index.type :date
+    index.as :stored_sortable
+  end
+
+  # TODO: Reel #: https://github.com/samvera-labs/uri_selection_wg/issues/2
+  # TODO: Titles on reel
+
+  # BasicMetadata must be included last
+  include ::Hyrax::BasicMetadata
+
+  # for GeoNames autocomplete lookup
+  include NewspaperWorks::PlaceOfPublicationBehavior
+
+  # relationship methods
+
+  def publication
+    result = member_of.select { |v| v.instance_of?(NewspaperTitle) }
+    result[0] unless result.empty?
+  end
+
+  def pages
+    members.select { |v| v.instance_of?(NewspaperPage) }
+  end
+end

data/app/models/newspaper_issue.rb

@@ -0,0 +1,115 @@
+# Newspaper Issue
+class NewspaperIssue < ActiveFedora::Base
+  # WorkBehavior mixes in minimal ::Hyrax::CoreMetadata fields of
+  # depositor, title, date_uploaded, and date_modified.
+  # https://samvera.github.io/customize-metadata-model.html#core-metadata
+  include ::Hyrax::WorkBehavior
+  include NewspaperWorks::NewspaperCoreMetadata
+
+  self.indexer = NewspaperIssueIndexer
+  # Change this to restrict which works can be added as a child.
+  # self.valid_child_concerns = []
+
+  # Validation and required fields:
+  validates :title, presence: {
+    message: 'Your work must have a title.'
+  }
+
+  validates_with NewspaperWorks::PublicationDateValidator
+
+  # TODO: Implement validations
+  # validates :resource_type, presence: {
+  #   message: 'A newspaper article requires a resource type.'
+  # }
+  # validates :language, presence: {
+  #   message: 'A newspaper article requires a language.'
+  # }
+  # validates :held_by, presence: {
+  #   message: 'A newspaper article requires a holding location.'
+  # }
+
+  # TODO: Reel #: https://github.com/samvera-labs/uri_selection_wg/issues/2
+
+  #  - Volume
+  property(
+    :volume,
+    predicate: ::RDF::Vocab::BIBO.volume,
+    multiple: false
+  ) do |index|
+    index.as :stored_searchable
+  end
+
+  # - Edition name
+  property(
+    :edition_name,
+    predicate: ::RDF::Vocab::BF2.editionStatement,
+    multiple: false
+  ) do |index|
+    index.as :stored_searchable
+  end
+
+  # - Edition number / enumeration
+  property(
+    :edition_number,
+    predicate: ::RDF::Vocab::BF2.editionEnumeration,
+    multiple: false
+  ) do |index|
+    index.as :stored_searchable
+  end
+
+  #  - Issue
+  property(
+    :issue_number,
+    predicate: ::RDF::Vocab::BIBO.issue,
+    multiple: false
+  ) do |index|
+    index.as :stored_searchable
+  end
+
+  # - Extent
+  property(
+    :extent,
+    predicate: ::RDF::Vocab::DC.extent,
+    multiple: false
+  ) do |index|
+    index.as :stored_searchable
+  end
+
+  #  - publication date
+  property(
+    :publication_date,
+    predicate: ::RDF::Vocab::DC.issued,
+    multiple: false
+  ) do |index|
+    index.type :date
+    index.as :stored_sortable
+  end
+
+  # BasicMetadata must be included last
+  include ::Hyrax::BasicMetadata
+
+  # for GeoNames autocomplete lookup
+  include NewspaperWorks::PlaceOfPublicationBehavior
+
+  # relationship methods
+  def publication
+    result = member_of.select { |v| v.instance_of?(NewspaperTitle) }
+    result[0] unless result.empty?
+  end
+
+  def articles
+    members.select { |v| v.instance_of?(NewspaperArticle) }
+  end
+
+  def pages
+    members.select { |v| v.instance_of?(NewspaperPage) }
+  end
+
+  def ordered_pages
+    ordered_members.to_a.select { |v| v.instance_of?(NewspaperPage) }
+  end
+
+  def ordered_page_ids
+    ordered_pages.map(&:id)
+  end
+end