iiif_print 1.0.0 → 2.0.0

data/app/transactions/hyrax/transactions/steps/delete_all_file_sets_decorator.rb

@@ -0,0 +1,35 @@
+# frozen_string_literal: true
+
+# OVERRIDE Hyrax 5.0.0rc2 to add file_set.iiif_print_conditionally_destroy_spawned_children with user args
+
+module Hyrax
+  module Transactions
+    module Steps
+      module DeleteAllFileSetsDecorator
+        include Dry::Monads[:result]
+
+        ##
+        # @param [Valkyrie::Resource] resource
+        # @param [::User] the user resposible for the delete action
+        #
+        # @return [Dry::Monads::Result]
+        def call(resource, user: nil)
+          return Failure(:resource_not_persisted) unless resource.persisted?
+
+          @query_service.custom_queries.find_child_file_sets(resource: resource).each do |file_set|
+            return Failure[:failed_to_delete_file_set, file_set] unless
+              Hyrax::Transactions::Container['file_set.destroy']
+              .with_step_args('file_set.remove_from_work' => { user: user },
+                              'file_set.delete' => { user: user },
+                              'file_set.iiif_print_conditionally_destroy_spawned_children' => { user: user })
+              .call(file_set).success?
+          rescue ::Ldp::Gone
+            nil
+          end
+
+          Success(resource)
+        end
+      end
+    end
+  end
+end

data/app/views/catalog/_index_header_list_default.html.erb

@@ -0,0 +1,13 @@
+<%# OVERRIDE Hyrax 2.9.6 to show parent_query params if metadata is found in parent record %>
+
+<div class="search-results-title-row">
+  <h3 class="search-result-title">
+  <% if params['q'].present? && document.any_highlighting? %>
+    <%= link_to document.title_or_label, [document, { parent_query: params['q'] }] %></h3>
+  <% elsif params['q'].present? %>
+    <%= link_to document.title_or_label, [document, { query: params['q'] }] %></h3>
+  <% else %>
+    <%= link_to document.title_or_label, document %></h3>
+  <% end %>
+  </h3>
+</div>

data/app/views/hyrax/base/_representative_media.html.erb

@@ -1,9 +1,10 @@
-<% if presenter.iiif_viewer? %>
-  <% if defined?(viewer) && viewer %>
+<% if presenter.representative_id.present? && presenter.representative_presenter.present? %>
+  <% if defined?(viewer) && viewer && presenter.iiif_viewer? %>
     <%= iiif_viewer_display presenter %>
   <% else %>
     <%= render media_display_partial(presenter.representative_presenter), file_set: presenter.representative_presenter %>
   <% end %>
 <% else %>
-  <%= image_tag 'default.png', class: "canonical-image", alt: 'default representative image' %>
+  <% alt = block_for(name: 'default_work_image_text') || 'Default work thumbnail' %>
+  <%= image_tag default_work_image, class: "canonical-image", alt: alt %>
 <% end %>

data/app/views/hyrax/base/iiif_viewers/_universal_viewer.html.erb

@@ -1,7 +1,7 @@
 <div class="viewer-wrapper">
   <iframe
     id="uv-iframe"
-    src="<%= universal_viewer_base_url %>#?manifest=<%= main_app.polymorphic_url [main_app, :manifest, presenter], { locale: nil } %>&config=<%= universal_viewer_config_url %>"
+    src="<%= universal_viewer_base_url %>#?manifest=<%= main_app.polymorphic_url [main_app, :manifest, presenter], { locale: nil } %>&config=<%= universal_viewer_config_url %><%= uv_search_param %>"
     allowfullscreen="true"
     frameborder="0"
   ></iframe>

data/app/views/hyrax/file_sets/_show_actions.html.erb

@@ -0,0 +1,24 @@
+<div class="form-actions">
+  <% if Hyrax.config.analytics? %>
+    <% # turbolinks needs to be turned off or the page will use the cache and the %>
+    <% # analytics graph will not show unless the page is refreshed. %>
+    <%= link_to t('.analytics'), @presenter.stats_path, id: 'stats', class: 'btn btn-default', data: { turbolinks: false } %>
+  <% end %>
+
+  <%# Hyrax 2.9.6 does not respond to workflow_restriction; that is something added in later versions. %>
+  <% if @presenter.editor? && (!respond_to?(:workflow_restriction?) || !workflow_restriction?(@presenter)) %>
+      <%= link_to t(".edit_this", type: @presenter.human_readable_type), edit_polymorphic_path([main_app, @presenter]),
+                  class: 'btn btn-default' %>
+      <%= link_to t(".delete_this", type: @presenter.human_readable_type), [main_app, @presenter],
+                  class: 'btn btn-danger', data: { confirm: t(".confirm_delete_this", type: @presenter.human_readable_type) },
+                  method: :delete %>
+  <% end %>
+
+  <% if @presenter.show_split_button? && @presenter.editor? && @presenter.pdf? %>
+    <%= link_to t("iiif_print.file_set.split_this"), iiif_print.split_pdf_path(@presenter),
+     class: 'btn btn-default', data: { confirm: t("iiif_print.file_set.confirm_split_this") },
+     method: :post %>
+  <% end %>
+
+  <%= render 'social_media' %>
+</div>

data/config/initializers/simple_schema_loader.rb

@@ -0,0 +1 @@
+"Hyrax::SimpleSchemaLoader".safe_constantize&.prepend(IiifPrint::SimpleSchemaLoaderDecorator)

data/config/locales/iiif_print.en.yml

@@ -52,6 +52,10 @@ en:
         label: 'Place of publication'
       publication_title:
         label: 'Publication'
+    file_set:
+      split_this: 'Re-Split PDF'
+      confirm_split_this: 'Re-Split PDF'
+      split_submitted: 'Submitted PDF re-splitting job for FileSet ID=%{id}'
     newspapers_search:
       title:                    'Search Newspapers'
       text:                     'Use this form to search for full-text newspaper content.'

data/config/metadata/child_works_from_pdf_splitting.yaml

@@ -0,0 +1,21 @@
+attributes:
+  is_child:
+    type: bool
+    multiple: false
+    index_keys:
+      - "is_child_bsi"
+    form:
+      required: false
+      primary: false
+      multiple: false
+    predicate: "http://id.loc.gov/vocabulary/identifiers/isChild"
+  split_from_pdf_id:
+    type: string
+    multiple: false
+    index_keys:
+      - "split_from_pdf_id_ssi"
+    form:
+      required: false
+      primary: false
+      multiple: false
+    predicate: "http://id.loc.gov/vocabulary/identifiers/splitFromPdfId"

data/config/routes.rb

@@ -0,0 +1,3 @@
+IiifPrint::Engine.routes.draw do
+  post "split_pdfs/:file_set_id" => "split_pdfs#create", as: :split_pdf
+end

data/db/migrate/20181214181358_create_iiif_print_derivative_attachments.rb

@@ -1,12 +1,14 @@
 class CreateIiifPrintDerivativeAttachments < ActiveRecord::Migration[5.0]
   def change
-    create_table :iiif_print_derivative_attachments do |t|
-      t.string :fileset_id
-      t.string :path
-      t.string :destination_name
+    unless table_exists?(:iiif_print_derivative_attachments)
+      create_table :iiif_print_derivative_attachments do |t|
+        t.string :fileset_id
+        t.string :path
+        t.string :destination_name
 
-      t.timestamps
+        t.timestamps
+      end
+      add_index :iiif_print_derivative_attachments, :fileset_id
     end
-    add_index :iiif_print_derivative_attachments, :fileset_id
   end
 end

data/db/migrate/20190107165909_create_iiif_print_ingest_file_relations.rb

@@ -1,11 +1,13 @@
 class CreateIiifPrintIngestFileRelations < ActiveRecord::Migration[5.0]
   def change
-    create_table :iiif_print_ingest_file_relations do |t|
-      t.string :file_path
-      t.string :derivative_path
+    unless table_exists?(:iiif_print_ingest_file_relations)
+      create_table :iiif_print_ingest_file_relations do |t|
+        t.string :file_path
+        t.string :derivative_path
 
-      t.timestamps
+        t.timestamps
+      end
+      add_index :iiif_print_ingest_file_relations, :file_path
     end
-    add_index :iiif_print_ingest_file_relations, :file_path
   end
 end

data/db/migrate/20230109000000_create_iiif_print_pending_relationships.rb

@@ -1,11 +1,13 @@
 class CreateIiifPrintPendingRelationships < ActiveRecord::Migration[5.1]
   def change
-    create_table :iiif_print_pending_relationships do |t|
-      t.string :child_title, null: false
-      t.string :parent_id, null: false
-      t.string :child_order, null: false
-      t.timestamps
+    unless table_exists?(:iiif_print_pending_relationships)
+      create_table :iiif_print_pending_relationships do |t|
+        t.string :child_title, null: false
+        t.string :parent_id, null: false
+        t.string :child_order, null: false
+        t.timestamps
+      end
+      add_index :iiif_print_pending_relationships, :parent_id
     end
-    add_index :iiif_print_pending_relationships, :parent_id
   end
 end

data/db/migrate/20231110163052_add_model_details_to_iiif_print_pending_relationships.rb

@@ -0,0 +1,7 @@
+class AddModelDetailsToIiifPrintPendingRelationships < ActiveRecord::Migration[5.2]
+  def change
+    add_column :iiif_print_pending_relationships, :parent_model, :string unless column_exists?(:iiif_print_pending_relationships, :parent_model)
+    add_column :iiif_print_pending_relationships, :child_model, :string  unless column_exists?(:iiif_print_pending_relationships, :child_model)
+    add_column :iiif_print_pending_relationships, :file_id, :string  unless column_exists?(:iiif_print_pending_relationships, :file_id)
+  end
+end

data/docker-compose.yml

@@ -85,12 +85,12 @@ services:
     environment:
       - VIRTUAL_PORT=3000
       - VIRTUAL_HOST=.hyku.test
-    command: tail -f /dev/null
+    # command: tail -f /dev/null
     ##
     ## Similar to the above, except we will bundle and then tell the container
     ## to wait.  You'll then need to bash into the web container to do much of
     ## anything.
-    # command: sh -l -c "bundle && echo \"Finished bundling now waiting...\" && tail -f /dev/null"
+    command: sh -l -c "bundle install && echo \"Finished bundling now waiting...\" && tail -f /dev/null"
     depends_on:
       db:
         condition: service_started

data/iiif_print.gemspec

@@ -12,21 +12,18 @@ Gem::Specification.new do |spec|
   spec.email       = ['sean.upton@utah.edu', 'jacob.reed@utah.edu',
                       'brian.mcbride@utah.edu', 'eenglish@bpl.org', 'kirk.wang@scientist.com',
                       'larita@scientist.com', 'jeremy.n.friesen@gmail.com']
-  spec.homepage    = 'https://github.com/samvera-labs/iiif_print'
+  spec.homepage    = 'https://github.com/scientist-softserv/iiif_print/'
   spec.description = 'Gem/Engine for IIIF Print works in Hyrax-based Samvera Application.'
   spec.summary     = <<-SUMMARY
-  iiif_print is a Rails Engine gem providing model and administrative
-  functions to Hyrax-based Samvera applications, for management of
-  (primarily scanned) content.
+  IiifPrint is a gem (Rails "engine") for Hyrax-based digital repository applications to support displaying parent/child works in the same viewer (Universal Viewer) and the ability to search OCR from the parent work to the child work(s). IiifPring was originally based off of the samvera-labs Newspaper gem.
 SUMMARY
   spec.license = 'Apache-2.0'
-  spec.files = `git ls-files`.split($OUTPUT_RECORD_SEPARATOR)
+  spec.files = `git ls-files`.split($OUTPUT_RECORD_SEPARATOR).select { |f| File.dirname(f) !~ %r{\A"?spec\/?} && f != 'bin/rails' }
   spec.test_files = spec.files.grep(%r{^(test|spec|features)/})
-  spec.add_dependency 'blacklight_iiif_search', '~> 1.0'
-  spec.add_dependency 'dry-monads', '~> 1.4.0'
-  spec.add_dependency 'hyrax', '>= 2.5', '< 4.0'
+  spec.add_dependency 'blacklight_iiif_search', '>= 1.0', '< 3.0'
+  spec.add_dependency 'derivative-rodeo', "~> 0.5"
+  spec.add_dependency 'hyrax', '>= 2.5', '< 6'
   spec.add_dependency 'nokogiri', '>=1.13.2'
-  spec.add_dependency 'rails', '~> 5.0'
   spec.add_dependency 'rdf-vocab', '~> 3.0'
 
   spec.add_development_dependency 'bixby'
@@ -34,10 +31,14 @@ SUMMARY
   spec.add_development_dependency 'engine_cart', '~> 2.2'
   spec.add_development_dependency "factory_bot", '~> 4.4'
   spec.add_development_dependency 'fcrepo_wrapper', '~> 0.5', '>= 0.5.1'
+  # TODO: We want to remove dependency on this
   spec.add_development_dependency 'newspaper_works_fixtures', '~> 0.3', '>=0.3.1'
   spec.add_development_dependency 'rails-controller-testing', '~> 1'
-  spec.add_development_dependency 'rspec-rails', '~> 3.1'
+  spec.add_development_dependency 'json-canonicalization', '0.3.1'
+  spec.add_development_dependency 'rspec-rails'
   spec.add_development_dependency 'rspec-activemodel-mocks'
   spec.add_development_dependency 'shoulda-matchers', '~> 3.1'
   spec.add_development_dependency 'solr_wrapper', '>= 1.1', '< 3.0'
+  spec.add_development_dependency 'solargraph'
+  spec.add_development_dependency 'yard'
 end

data/lib/generators/iiif_print/install_generator.rb

@@ -15,13 +15,21 @@ module IiifPrint
       say_status('info',
                  'BLACKLIGHT IIIF SEARCH NOT INSTALLED; INSTALLING BLACKLIGHT IIIF SEARCH',
                  :blue)
-      generate 'blacklight_iiif_search:install'
+      generate 'blacklight_iiif_search:install --skip-solr'
     end
 
     def catalog_controller_configuration
       generate 'iiif_print:catalog_controller'
     end
 
+    def install_routes
+      return if IO.read('config/routes.rb').include?('mount IiifPrint::Engine')
+
+      inject_into_file 'config/routes.rb', after: /mount Hyrax::Engine\s*\n/ do
+        "  mount IiifPrint::Engine, at: '/'\n"\
+      end
+    end
+
     def inject_configuration
       copy_file 'config/initializers/iiif_print.rb'
     end
@@ -30,6 +38,10 @@ module IiifPrint
       generate 'iiif_print:assets'
     end
 
+    def inject_helper
+      copy_file 'helpers/iiif_print_helper.rb' 'app/helpers/iiif_print_helper.rb'
+    end
+
     # Blacklight IIIF Search generator has some linting that does not agree with CircleCI on Hyku
     # ref https://github.com/boston-library/blacklight_iiif_search/blob/v1.0.0/lib/generators/blacklight_iiif_search/controller_generator.rb
     # the follow two methods does a clean up to appease Rubocop
@@ -48,5 +60,13 @@ module IiifPrint
       contents.insert(0, "# frozen_string_literal: true\n\n")
       File.write(file, contents)
     end
+
+    def add_allinson_flex_fields_method_to_iiif_search_builder
+      file_path = "app/models/iiif_search_builder.rb"
+      contents = File.read(file_path)
+      contents.gsub!('include Blacklight::Solr::SearchBuilderBehavior', "include Blacklight::Solr::SearchBuilderBehavior\n  include IiifPrint::AllinsonFlexFields")
+      contents.gsub!('self.default_processor_chain += [:ocr_search_params]', 'self.default_processor_chain += %i[ocr_search_params include_allinson_flex_fields]')
+      File.write(file_path, contents)
+    end
   end
 end

data/lib/generators/iiif_print/templates/config/initializers/iiif_print.rb

@@ -1,3 +1,4 @@
+# rubocop:disable Lint/UnusedBlockArgument
 IiifPrint.config do |config|
   # NOTE: WorkTypes and models are used synonymously here.
   # Add models to be excluded from search so the user
@@ -15,8 +16,14 @@ IiifPrint.config do |config|
   # @example
   #   config.excluded_model_name_solr_field_key = 'some_solr_field_key'
 
-  # Configure how the manifest sorts the canvases, by default it sorts by :title,
-  # but a different model property may be desired such as :date_published
-  # @example
-  #   config.sort_iiif_manifest_canvases_by = :date_published
+  if Rails.env.development?
+    if DerivativeRodeo.config.aws_s3_access_key_id.present? && DerivativeRodeo.config.aws_s3_secret_access_key.present?
+      Rails.logger.info("DerivativeRodeo S3 Credentials detected using 's3' for IiifPrint::DerivativeRodeoService.preprocessed_location_adapter_name")
+      IiifPrint::DerivativeRodeoService.preprocessed_location_adapter_name = 's3'
+    else
+      Rails.logger.info("DerivativeRodeo S3 Credentials not-detected using 'file' for IiifPrint::DerivativeRodeoService.preprocessed_location_adapter_name")
+      IiifPrint::DerivativeRodeoService.preprocessed_location_adapter_name = 'file'
+    end
+  end
 end
+# rubocop:enable Lint/UnusedBlockArgument

data/lib/generators/iiif_print/templates/helpers/iiif_print_helper.rb

@@ -0,0 +1,5 @@
+# frozen_string_literal: true
+
+module IiifPrintHelper
+  include IiifPrint::IiifPrintHelperBehavior
+end

data/lib/iiif_print/base_derivative_service.rb

@@ -7,7 +7,11 @@ module IiifPrint
     class_attribute :target_extension, default: nil
 
     def initialize(file_set)
-      @file_set = file_set
+      @file_set = if file_set.is_a?(Hyrax::FileMetadata)
+                    Hyrax.query_service.find_by(id: file_set.file_set_id)
+                  else
+                    file_set
+                  end
       @dest_path = nil
       @source_path = nil
       @source_meta = nil
@@ -25,7 +29,11 @@ module IiifPrint
     # @see IiifPrint::PluggableDerivativeService#plugins_for
     # @return [Boolean]
     def valid?
-      true
+      # @note We are taking a shortcut because currently we are only concerned about images.
+      # @TODO: verify if this works for ActiveFedora and if so, remove commented code.
+      #        If not, modify to use adapter.
+      # file_set.class.image_mime_types.include?(file_set.mime_type)
+      file_set.original_file.image?
     end
 
     def derivative_path_factory
@@ -109,5 +117,9 @@ module IiifPrint
       # intermediate -> PDF
       im_convert
     end
+
+    def mime_type_for(extension)
+      Marcel::MimeType.for extension: extension
+    end
   end
 end

data/lib/iiif_print/blacklight_iiif_search/annotation_decorator.rb

@@ -2,6 +2,7 @@
 module IiifPrint
   module BlacklightIiifSearch
     module AnnotationDecorator
+      INVALID_MATCH_TEXT = "#xywh=INVALID,INVALID,INVALID,INVALID".freeze
       ##
       # Create a URL for the annotation
       # use a Hyrax-y URL syntax:
@@ -28,23 +29,33 @@ module IiifPrint
       # @return [String]
       def coordinates
         return default_coords if query.blank?
+
+        sanitized_query = sanitize_query.downcase
         coords_json = fetch_and_parse_coords
-        return default_coords unless coords_json && coords_json['coords']
-        query_terms = query.split(' ').map(&:downcase)
+        return derived_coords_json_and_properties(sanitized_query) unless coords_json && coords_json['coords']
+
+        query_terms = sanitized_query.split(' ')
+
         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
+        return default_coords unless coords_array
+
         "#xywh=#{coords_array.join(',')}"
       end
 
+      def sanitize_query
+        query.match(additional_query_terms_regex)[1].strip
+      end
+
       ##
       # return the JSON word-coordinates file contents
       # @return [JSON]
       def fetch_and_parse_coords
-        coords = IiifPrint::Data::WorkDerivatives.data(from: file_set_id, of_type: 'json')
+        coords = IiifPrint.config.ocr_coords_from_json_function.call(file_set_id: file_set_id, document: document)
         return nil if coords.blank?
         begin
           JSON.parse(coords)
@@ -53,6 +64,23 @@ module IiifPrint
         end
       end
 
+      # This is a bit hacky but it is checking if any of the properties contain the query term
+      # if there are no coords and there is a metadata property match
+      # then we return the default coords
+      # else we insert a invalid match text to be stripped out at a later point
+      # @see IiifPrint::IiifSearchResponseDecorator#annotation_list
+      def derived_coords_json_and_properties(sanitized_query)
+        property = @document.keys.detect do |key|
+          (key.ends_with?("_tesim") || key.ends_with?("_tsim")) && property_includes_sanitized_query?(key, sanitized_query)
+        end
+
+        property ? default_coords : INVALID_MATCH_TEXT
+      end
+
+      def property_includes_sanitized_query?(property, sanitized_query)
+        @document[property].join.downcase.include?(sanitized_query)
+      end
+
       ##
       # a default set of coordinates
       # @return [String]
@@ -75,9 +103,33 @@ module IiifPrint
       def file_set_id
         return document['id'] if document.file_set?
 
-        file_set_ids = document['file_set_ids_ssim']
+        file_set_ids = document['member_ids_ssim']
         raise "#{self.class}: NO FILE SET ID" if file_set_ids.blank?
-        file_set_ids.first
+
+        # Since a parent work's `member_ids_ssim` can contain child work ids as well as file set ids,
+        # this will ensure that the file set id is indeed a `FileSet`
+        file_set_ids.detect { |id| SolrDocument.find(id).file_set? }
+      end
+
+      ##
+      # This method is a workaround to compensate for overriding the solr_params method in
+      # BlacklightIiifSearch::IiifSearch. In the override, the solr_params method adds an additional filter to the query
+      # to include either the object_relation_field OR the parent document's id and removes the :f parameter from the
+      # query. This resulted in the query split here returning more than the actual query term.
+      #
+      # @see IiifPrint::IiifSearchDecorator#solr_params
+      # @return [Regexp] A regular expression to find the last AND and everything after it
+      # @example
+      #   'foo AND (is_page_of_ssim:\"123123\" OR id:\"123123\")' #=> 'foo'
+      def additional_query_terms_regex
+        /(.*)(?= AND (\(.+\)|\w+)$)/
+      end
+
+      ##
+      # @return [IIIF::Presentation::Resource]
+      def text_resource_for_annotation
+        IIIF::Presentation::Resource.new('@type' => 'cnt:ContentAsText',
+                                         'chars' => sanitize_query)
       end
     end
   end

data/lib/iiif_print/catalog_search_builder.rb

@@ -11,19 +11,23 @@ module IiifPrint
     include IiifPrint::HighlightSearchParams
     # TODO: Do we need the following as a module?  It hides the behavior
     include IiifPrint::ExcludeModels
+    include IiifPrint::AllinsonFlexFields
 
     # NOTE: If you are using advanced_search, the :exclude_models and :highlight_search_params must
     # be added after the advanced_search methods (which are not part of this gem).  In other tests,
     # we found that having the advanced search processing after the two aforementioned processors
     # resulted in improper evaluation of keyword querying.
-    self.default_processor_chain += [:exclude_models, :highlight_search_params, :show_parents_only]
+    self.default_processor_chain += [:exclude_models,
+                                     :highlight_search_params,
+                                     :show_parents_only,
+                                     :include_allinson_flex_fields]
 
     # rubocop:enable Naming/PredicateName
     def show_parents_only(solr_parameters)
       query = if blacklight_params["include_child_works"] == 'true'
-                ActiveFedora::SolrQueryBuilder.construct_query(is_child_bsi: 'true')
+                IiifPrint.solr_construct_query(is_child_bsi: 'true')
               else
-                ActiveFedora::SolrQueryBuilder.construct_query(is_child_bsi: nil)
+                IiifPrint.solr_construct_query(is_child_bsi: nil)
               end
       solr_parameters[:fq] += [query]
     end