iiif_print 1.0.0

data/docker-compose.yml

@@ -0,0 +1,129 @@
+version: '3.8'
+
+x-app: &app
+  build:
+    context: .
+    target: hyku-base
+    args:
+      - EXTRA_APK_PACKAGES=less vim bash openjdk11-jre ffmpeg rsync exiftool
+  # image: ghcr.io/scientist-softserv/palni-palci:${TAG:-latest}
+  env_file:
+    - .env
+  # NOTE: all common env variables moved to .env
+  volumes:
+    - node_modules:/app/samvera/hyrax-webapp/node_modules:cached
+    - uploads:/app/samvera/hyrax-webapp/public/uploads:cached
+    - assets:/app/samvera/hyrax-webapp/public/assets:cached
+    - cache:/app/samvera/hyrax-webapp/tmp/cache:cached
+    - .:/app/samvera/hyrax-webapp
+  networks:
+    internal:
+
+volumes:
+  assets:
+  cache:
+  db:
+  fcrepo:
+  node_modules:
+  redis:
+  solr:
+  uploads:
+  zk:
+  zoo:
+
+networks:
+  internal:
+
+services:
+  solr:
+    image: solr:7.7.2
+    environment:
+      - VIRTUAL_PORT=8983
+      - VIRTUAL_HOST=solr.iiif_print.test
+    command:
+      - sh
+      - "-c"
+      - "precreate-core hyrax_test /opt/solr/server/configsets/hyraxconf; solr-precreate hyrax /opt/solr/server/configsets/hyraxconf"
+    volumes:
+      - solr:/var/solr/data:cached
+      - ./spec/test_app_templates/solr/conf/:/opt/solr/server/configsets/hyraxconf
+    networks:
+      - internal
+
+  fcrepo:
+    image: ghcr.io/samvera/fcrepo4:4.7.5
+    volumes:
+      - fcrepo:/data:cached
+    env_file:
+      - .env
+    environment:
+      - VIRTUAL_PORT=8080
+      - VIRTUAL_HOST=fcrepo.hyku.test
+      - JAVA_OPTS=${JAVA_OPTS} -Dfcrepo.modeshape.configuration="classpath:/config/file-simple/repository.json" -Dfcrepo.object.directory="/data/objects" -Dfcrepo.binary.directory="/data/binaries"
+    expose:
+      - 8080
+    networks:
+      internal:
+
+  db:
+    image: postgres:11.1
+    env_file:
+      - .env
+    environment:
+      - POSTGRES_DB=${DB_NAME}
+      - POSTGRES_PASSWORD=${DB_PASSWORD}
+      - POSTGRES_USER=${DB_USER}
+      - VIRTUAL_PORT=5432
+      - VIRTUAL_HOST=db.hyku.test
+    volumes:
+      - db:/var/lib/postgresql/data
+    networks:
+      internal:
+
+  web:
+    <<: *app
+    environment:
+      - VIRTUAL_PORT=3000
+      - VIRTUAL_HOST=.hyku.test
+    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"
+    depends_on:
+      db:
+        condition: service_started
+      solr:
+        condition: service_started
+      fcrepo:
+        condition: service_started
+      redis:
+        condition: service_started
+      chrome:
+        condition: service_started
+    expose:
+      - 3000
+
+  redis:
+    image: redis:5
+    command: redis-server
+    volumes:
+      - redis:/data
+    networks:
+      internal:
+
+  chrome:
+    # password is 'secret'
+    image: selenium/standalone-chrome-debug:3.141.59-20201010
+    logging:
+      driver: none
+    volumes:
+      - /dev/shm:/dev/shm
+    shm_size: 3G
+    networks:
+      internal:
+    environment:
+      - JAVA_OPTS=-Dwebdriver.chrome.whitelistedIps=
+      - VIRTUAL_PORT=7900
+      - VIRTUAL_HOST=chrome.hyku.test

data/iiif_print.gemspec

@@ -0,0 +1,43 @@
+$LOAD_PATH.push File.expand_path('../lib', __FILE__)
+
+# version updated in one place:
+require 'iiif_print/version'
+
+# Gem description:
+Gem::Specification.new do |spec|
+  spec.name        = 'iiif_print'
+  spec.version     = IiifPrint::VERSION
+  spec.authors     = ['Sean Upton', 'Jacob Reed', 'Brian McBride',
+                      'Eben English', 'Kirk Wang', 'LaRita Robinson', 'Jeremy Friesen']
+  spec.email       = ['sean.upton@utah.edu', 'jacob.reed@utah.edu',
+                      'brian.mcbride@utah.edu', 'eenglish@bpl.org', 'kirk.wang@scientist.com',
+                      'larita@scientist.com', 'jeremy.n.friesen@gmail.com']
+  spec.homepage    = 'https://github.com/samvera-labs/iiif_print'
+  spec.description = 'Gem/Engine for IIIF Print works in Hyrax-based Samvera Application.'
+  spec.summary     = <<-SUMMARY
+  iiif_print is a Rails Engine gem providing model and administrative
+  functions to Hyrax-based Samvera applications, for management of
+  (primarily scanned) content.
+SUMMARY
+  spec.license = 'Apache-2.0'
+  spec.files = `git ls-files`.split($OUTPUT_RECORD_SEPARATOR)
+  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 'nokogiri', '>=1.13.2'
+  spec.add_dependency 'rails', '~> 5.0'
+  spec.add_dependency 'rdf-vocab', '~> 3.0'
+
+  spec.add_development_dependency 'bixby'
+  spec.add_development_dependency 'database_cleaner', '~> 1.3'
+  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'
+  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 'rspec-activemodel-mocks'
+  spec.add_development_dependency 'shoulda-matchers', '~> 3.1'
+  spec.add_development_dependency 'solr_wrapper', '>= 1.1', '< 3.0'
+end

data/lib/generators/iiif_print/assets_generator.rb

@@ -0,0 +1,29 @@
+require 'rails/generators'
+
+module IiifPrint
+  class AssetsGenerator < Rails::Generators::Base
+    desc "This generator installs the iiif_print CSS assets into your application"
+
+    source_root File.expand_path('../templates', __FILE__)
+
+    def inject_css
+      copy_file "iiif_print.scss", "app/assets/stylesheets/iiif_print.scss"
+    end
+
+    def inject_js
+      return if iiif_print_js_installed?
+      insert_into_file 'app/assets/javascripts/application.js', after: '//= require hyrax' do
+        <<-JS.strip_heredoc
+
+        //= require iiif_print
+        JS
+      end
+    end
+
+    private
+
+    def iiif_print_js_installed?
+      IO.read("app/assets/javascripts/application.js").include?('iiif_print')
+    end
+  end
+end

data/lib/generators/iiif_print/catalog_controller_generator.rb

@@ -0,0 +1,32 @@
+# adds controller-scope behavior to the implementing application
+require 'rails/generators'
+
+module IiifPrint
+  class CatalogControllerGenerator < Rails::Generators::Base
+    desc "
+      This generator makes the following changes to your app:
+      1. Adds index fields in CatalogController
+      2. Adjusts Blacklight IIIF Search configuration settings in CatalogController
+      "
+
+    def add_index_fields_to_catalog_controller
+      marker = 'configure_blacklight do |config|'
+      inject_into_file 'app/controllers/catalog_controller.rb', after: marker do
+        "\n\n    # IiifPrint index fields\n"\
+        "    config.add_index_field 'all_text_tsimv', highlight: true, helper_method: :render_ocr_snippets\n"
+      end
+    end
+
+    def adjust_catalog_controller_all_text_config
+      gsub_file('app/controllers/catalog_controller.rb',
+                " full_text_field: 'text',",
+                " full_text_field: 'all_text_tsimv',")
+    end
+
+    def adjust_catalog_controller_is_page_of_config
+      gsub_file('app/controllers/catalog_controller.rb',
+                " object_relation_field: 'is_page_of_s',",
+                " object_relation_field: 'is_page_of_ssim',")
+    end
+  end
+end

data/lib/generators/iiif_print/install_generator.rb

@@ -0,0 +1,52 @@
+require 'rails/generators'
+
+module IiifPrint
+  # Install Generator Class
+  # rubocop:disable Metrics/ClassLength
+  class InstallGenerator < Rails::Generators::Base
+    source_root File.expand_path('../templates', __FILE__)
+
+    def copy_migrations
+      rake "iiif_print:install:migrations"
+    end
+
+    def verify_biiif_installed
+      return if IO.read('app/controllers/catalog_controller.rb').include?('include BlacklightIiifSearch::Controller')
+      say_status('info',
+                 'BLACKLIGHT IIIF SEARCH NOT INSTALLED; INSTALLING BLACKLIGHT IIIF SEARCH',
+                 :blue)
+      generate 'blacklight_iiif_search:install'
+    end
+
+    def catalog_controller_configuration
+      generate 'iiif_print:catalog_controller'
+    end
+
+    def inject_configuration
+      copy_file 'config/initializers/iiif_print.rb'
+    end
+
+    def inject_assets
+      generate 'iiif_print:assets'
+    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
+    def lint_catalog_controller
+      file = "app/controllers/catalog_controller.rb"
+      contents = File.read(file)
+      contents.gsub!(/\n\s*\n\s*# IiifPrint index fields/, "\n    # IiifPrint index fields")
+      contents.gsub!(/\n\s*\n\s*# configuration for Blacklight IIIF Content Search/, "\n\n    # configuration for Blacklight IIIF Content Search")
+      File.write(file, contents)
+    end
+
+    # ref https://github.com/boston-library/blacklight_iiif_search/blob/v1.0.0/lib/generators/blacklight_iiif_search/templates/iiif_search_builder.rb
+    def lint_iiif_search_builder
+      file = "app/models/iiif_search_builder.rb"
+      contents = File.read(file)
+      contents.insert(0, "# frozen_string_literal: true\n\n")
+      File.write(file, contents)
+    end
+  end
+end

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

@@ -0,0 +1,22 @@
+IiifPrint.config do |config|
+  # NOTE: WorkTypes and models are used synonymously here.
+  # Add models to be excluded from search so the user
+  # would not see them in the search results.
+  # by default, use the human readable versions like:
+  # @example
+  #   # config.excluded_model_name_solr_field_values = ['Generic Work', 'Image']
+  #
+  # config.excluded_model_name_solr_field_values = []
+
+  # Add configurable solr field key for searching,
+  # default key is: 'human_readable_type_sim'
+  # if another key is used, make sure to adjust the
+  # config.excluded_model_name_solr_field_values to match
+  # @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
+end

data/lib/generators/iiif_print/templates/iiif_print.scss

@@ -0,0 +1 @@
+@import 'iiif_print/iiif_print';

data/lib/iiif_print/base_derivative_service.rb

@@ -0,0 +1,113 @@
+module IiifPrint
+  # Base type for IiifPrint derivative services
+  class BaseDerivativeService
+    attr_reader :file_set, :master_format
+    delegate :uri, to: :file_set
+
+    class_attribute :target_extension, default: nil
+
+    def initialize(file_set)
+      @file_set = file_set
+      @dest_path = nil
+      @source_path = nil
+      @source_meta = nil
+    end
+
+    ##
+    # We assume that for the file set's parent that this is an acceptable plugin.  Now, we ask for
+    # this specific file_set is it valid.  For example, we would not attempt to extract text from a
+    # movie even though the parent work says to attempt to extract text on any attached file sets.
+    # Put another way, we can upload a PDF or a Movie to the parent.
+    #
+    # In subclass, you'll want to consider the attributes of the file_set and whether that subclass
+    # should process the given file_set.
+    #
+    # @see IiifPrint::PluggableDerivativeService#plugins_for
+    # @return [Boolean]
+    def valid?
+      true
+    end
+
+    def derivative_path_factory
+      Hyrax::DerivativePath
+    end
+
+    # prepare full path for passed extension/destination name, return path
+    def prepare_path(extension)
+      dest_path = derivative_path_factory.derivative_path_for_reference(
+        @file_set,
+        extension
+      )
+      dir = File.join(dest_path.split('/')[0..-2])
+      FileUtils.mkdir_p(dir) unless Dir.exist?(dir)
+      dest_path
+    end
+
+    # calculate and ensure directory components for singular @dest_path
+    #   should only be used by subclasses producing a single derivative
+    def load_destpath
+      @dest_path = prepare_path(target_extension)
+    end
+
+    def identify
+      return @source_meta unless @source_meta.nil?
+      @source_meta = IiifPrint::ImageTool.new(@source_path).metadata
+    end
+
+    def mime_type
+      identify[:content_type]
+    end
+
+    def use_color?
+      identify[:color] == 'color'
+    end
+
+    # is source one-bit monochrome?
+    def one_bit?
+      identify[:color] == 'monochrome'
+    end
+
+    def create_derivatives(filename)
+      # presuming that filename is full path to source file
+      @source_path = filename
+
+      # Get destination path from Hyrax for file extension defined in
+      #   self.target_extension constant on respective derivative service subclass.
+      load_destpath
+    end
+
+    def cleanup_derivatives(extension = target_extension, *_args)
+      derivative_path_factory.derivatives_for_reference(file_set).each do |path|
+        FileUtils.rm_f(path) if path.ends_with?(extension)
+      end
+    end
+
+    def jp2_to_intermediate
+      intermediate_path = File.join(Dir.mktmpdir, 'intermediate.tif')
+      jp2_cmd = "opj_decompress -i #{@source_path} -o #{intermediate_path}"
+      # make intermediate, then...
+      `#{jp2_cmd}`
+      intermediate_path
+    end
+
+    def convert_cmd
+      raise NotImplementedError, 'Calling subclass missing convert_cmd method'
+    end
+
+    # convert non-JP2 source/primary file to PDF derivative with ImageMagick6
+    #   calls convert_cmd on calling subclasses
+    def im_convert
+      `#{convert_cmd}`
+    end
+
+    # convert JP2 source/primary file to PDF derivative, via
+    #   opj_decompress to intermediate TIFF, then ImageMagick6 convert
+    def jp2_convert
+      # jp2 source -> intermediate
+      intermediate_path = jp2_to_intermediate
+      @source_path = intermediate_path
+      # intermediate -> PDF
+      im_convert
+    end
+  end
+end

data/lib/iiif_print/blacklight_iiif_search/annotation_decorator.rb

@@ -0,0 +1,84 @@
+# customize behavior for IiifSearch
+module IiifPrint
+  module BlacklightIiifSearch
+    module AnnotationDecorator
+      ##
+      # 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 = IiifPrint::Data::WorkDerivatives.data(from: file_set_id, of_type: '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
+        return document['id'] if document.file_set?
+
+        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/lib/iiif_print/catalog_search_builder.rb

@@ -0,0 +1,31 @@
+require 'hyrax/catalog_search_builder'
+
+module IiifPrint
+  # This class extends the base Hyrax::CatalogSearchBuilder by:
+  #
+  # - supporting highlighting of snippets in results
+  # - excluding models from search result; with complex works you might want to skip some of those
+  #   works.
+  class CatalogSearchBuilder < Hyrax::CatalogSearchBuilder
+    # TODO: Do we need the following as a module?  It hides the behavior
+    include IiifPrint::HighlightSearchParams
+    # TODO: Do we need the following as a module?  It hides the behavior
+    include IiifPrint::ExcludeModels
+
+    # 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]
+
+    # 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')
+              else
+                ActiveFedora::SolrQueryBuilder.construct_query(is_child_bsi: nil)
+              end
+      solr_parameters[:fq] += [query]
+    end
+  end
+end

data/lib/iiif_print/configuration.rb

@@ -0,0 +1,99 @@
+module IiifPrint
+  class Configuration
+    attr_writer :after_create_fileset_handler
+
+    # @param file_set [FileSet]
+    # @param user [User]
+    def handle_after_create_fileset(file_set, user)
+      if defined? @after_create_fileset_handler
+        @after_create_fileset_handler.call(file_set, user)
+      else
+        IiifPrint::Data.handle_after_create_fileset(file_set, user)
+      end
+    end
+
+    attr_writer :excluded_model_name_solr_field_values
+    # By default, this uses an array of human readable types
+    #   ex: ['Generic Work', 'Image']
+    # @return [Array<String>]
+    def excluded_model_name_solr_field_values
+      return @excluded_model_name_solr_field_values unless @excluded_model_name_solr_field_values.nil?
+      @excluded_model_name_solr_field_values = []
+    end
+
+    # This method wraps Hyrax's configuration so we can sniff out the correct method to use.  The
+    # {Hyrax::Configuration#whitelisted_ingest_dirs} is deprecated in favor of
+    # {Hyrax::Configuration#registered_ingest_dirs}.
+    #
+    # @return [Array<String>]
+    def registered_ingest_dirs
+      if Hyrax.config.respond_to?(:registered_ingest_dirs)
+        Hyrax.config.registered_ingest_dirs
+      else
+        Hyrax.config.whitelisted_ingest_dirs
+      end
+    end
+
+    attr_writer :excluded_model_name_solr_field_key
+    # A string of a solr field key
+    # @return [String]
+    def excluded_model_name_solr_field_key
+      return "human_readable_type_sim" unless defined?(@excluded_model_name_solr_field_key)
+      @excluded_model_name_solr_field_key
+    end
+
+    attr_writer :default_iiif_manifest_version
+    def default_iiif_manifest_version
+      @default_iiif_manifest_version || 2
+    end
+
+    attr_writer :metadata_fields
+    # rubocop:disable Metrics/MethodLength
+    # @api private
+    # @note These fields will appear in rendering order.
+    # @todo To move this to an `@api public` state, we need to consider what a proper configuration looks like.
+    def metadata_fields
+      @metadata_fields ||= {
+        title: {},
+        description: {},
+        collection: {},
+        abstract: {},
+        date_modified: {},
+        creator: { render_as: :faceted },
+        contributor: { render_as: :faceted },
+        subject: { render_as: :faceted },
+        publisher: { render_as: :faceted },
+        language: { render_as: :faceted },
+        identifier: { render_as: :linked },
+        keyword: { render_as: :faceted },
+        date_created: { render_as: :linked },
+        based_near_label: {},
+        related_url: { render_as: :external_link },
+        resource_type: { render_as: :faceted },
+        source: {},
+        extent: {},
+        rights_statement: { render_as: :rights_statement },
+        rights_notes: {},
+        access_right: {},
+        license: { render_as: :license },
+        searchable_text: {}
+      }
+    end
+    # rubocop:enable Metrics/MethodLength
+
+    attr_writer :sort_iiif_manifest_canvases_by
+    def sort_iiif_manifest_canvases_by
+      @sort_iiif_manifest_canvases_by || :title
+    end
+
+    attr_writer :additional_tessearct_options
+    ##
+    # The additional options to pass to the Tesseract configuration
+    #
+    # @see https://tesseract-ocr.github.io/tessdoc/Command-Line-Usage.html
+    # @return [String]
+    def additional_tessearct_options
+      @additional_tessearct_options || ""
+    end
+  end
+end

data/lib/iiif_print/data/fileset_helper.rb

@@ -0,0 +1,25 @@
+module IiifPrint
+  module Data
+    # Mixin module for fileset methods for work, presumes an @work
+    #   instance attribute refering to a work object
+    module FilesetHelper
+      def fileset_id
+        # if context is itself a string, presume it is a file set id
+        return @work if @work.is_a? String
+        # if context is not a String, presume a work or fileset context:
+        fileset.nil? ? nil : fileset.id
+      end
+
+      def first_fileset
+        # if context is fileset id (e.g. caller is view partial) string,
+        #   get the fileset from that id
+        return FileSet.find(@work) if @work.is_a?(String)
+        # if "work" context is a FileSet, not actual work, return it
+        return @work if @work.is_a? FileSet
+        # in most cases, get from work's members:
+        filesets = @work.members.select { |m| m.is_a? FileSet }
+        filesets.empty? ? nil : filesets[0]
+      end
+    end
+  end
+end