newspaper_works 0.1.0

data/lib/generators/newspaper_works/catalog_controller_generator.rb

@@ -0,0 +1,60 @@
+# adds controller-scope behavior to the implementing application
+require 'rails/generators'
+
+module NewspaperWorks
+  class CatalogControllerGenerator < Rails::Generators::Base
+    source_root File.expand_path('../templates', __FILE__)
+
+    desc "
+  This generator makes the following changes to your app:
+   1. Adds index fields in CatalogController
+   2. Adds facet fields in CatalogController
+   3. Adds sort fields 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    # NewspaperWorks index fields\n"\
+        "    config.add_index_field 'publication_title_ssi', label: I18n.t('newspaper_works.attributes.publication_title.label'), link_to_search: 'publication_title_ssi'\n"\
+        "    config.add_index_field solr_name('publication_date', :stored_sortable, type: :date), label: 'Publication date', helper_method: :human_readable_date\n"\
+        "    config.add_index_field solr_name('place_of_publication_label', :stored_searchable), label: I18n.t('newspaper_works.attributes.place_of_publication.label'), link_to_search: solr_name('place_of_publication_label', :facetable)\n"\
+        "    config.add_index_field solr_name('publication_date_start', :stored_sortable, type: :date), label: 'Publication date (start)', helper_method: :human_readable_date\n"\
+        "    config.add_index_field solr_name('publication_date_end', :stored_sortable, type: :date), label: 'Publication date (end)', helper_method: :human_readable_date\n"\
+        "    config.add_index_field 'all_text_tsimv', highlight: true, helper_method: :render_ocr_snippets\n"
+      end
+    end
+
+    def add_facets_to_catalog_controller
+      marker = 'configure_blacklight do |config|'
+      inject_into_file 'app/controllers/catalog_controller.rb', after: marker do
+        "\n\n    # NewspaperWorks facet fields\n"\
+        "    config.add_facet_field solr_name('place_of_publication_city', :facetable), label: 'Place of publication', limit: 5\n"\
+        "    config.add_facet_field 'publication_title_ssi', label: 'Publication title', limit: 5\n"\
+        "    config.add_facet_field solr_name('genre', :facetable), label: 'Article type', limit: 5\n\n"\
+        "    # additional NewspaperWorks fields not displayed in the facet list,\n"\
+        "    # but below definitions give labels to filters for linked metadata\n"\
+        "    config.add_facet_field solr_name('place_of_publication_label', :facetable), label: 'Place of publication', if: false\n"\
+        "    config.add_facet_field solr_name('issn', :facetable), label: 'ISSN', if: false\n"\
+        "    config.add_facet_field solr_name('lccn', :facetable), label: 'LCCN', if: false\n"\
+        "    config.add_facet_field solr_name('oclcnum', :facetable), label: 'OCLC #', if: false\n"\
+        "    config.add_facet_field solr_name('held_by', :facetable), label: 'Held by', if: false\n"\
+        "    config.add_facet_field solr_name('author', :facetable), label: 'Author', if: false\n"\
+        "    config.add_facet_field solr_name('photographer', :facetable), label: 'Photographer', if: false\n"\
+        "    config.add_facet_field solr_name('geographic_coverage', :facetable), label: 'Geographic coverage', if: false\n"\
+        "    config.add_facet_field solr_name('preceded_by', :facetable), label: 'Preceded by', if: false\n"\
+        "    config.add_facet_field solr_name('succeeded_by', :facetable), label: 'Succeeded by', if: false\n"\
+        "    config.add_facet_field 'first_page_bsi', label: 'First page', if: false\n"
+      end
+    end
+
+    def add_pubdate_sort_to_catalog_controller
+      marker = 'config.add_sort_field "#{modified_field} asc", label: "date modified \u25B2"'
+      inject_into_file 'app/controllers/catalog_controller.rb', after: marker do
+        "\n\n    # NewspaperWorks sort fields\n"\
+        '    config.add_sort_field "publication_date_dtsi desc", label: "publication date \u25BC"
+    config.add_sort_field "publication_date_dtsi asc", label: "publication date \u25B2"'
+      end
+    end
+  end
+end

data/lib/generators/newspaper_works/install_generator.rb

@@ -0,0 +1,97 @@
+require 'rails/generators'
+
+module NewspaperWorks
+  # Install Generator Class
+  class InstallGenerator < Rails::Generators::Base
+    source_root File.expand_path('../templates', __FILE__)
+
+    def copy_migrations
+      rake "newspaper_works:install:migrations"
+    end
+
+    def register_worktypes
+      inject_into_file 'config/initializers/hyrax.rb',
+                       after: "Hyrax.config do |config|\n" do
+        "  # Configuration generated by `rails g newspaper_works:install`\n" \
+          "  config.register_curation_concern :newspaper_article\n" \
+          "  config.register_curation_concern :newspaper_container\n" \
+          "  config.register_curation_concern :newspaper_issue\n" \
+          "  config.register_curation_concern :newspaper_page\n" \
+          "  config.register_curation_concern :newspaper_title\n" \
+          "  config.callback.set(:after_create_fileset) do |file_set, user|\n" \
+          "    require 'newspaper_works'\n" \
+          "    NewspaperWorks::Data.handle_after_create_fileset(file_set, user)\n" \
+          "  end\n" \
+          "  #== END GENERATED newspaper_works CONFIG ==\n\n"
+      end
+    end
+
+    def inject_routes
+      inject_into_file 'config/routes.rb',
+                       after: "Rails.application.routes.draw do\n" do
+        "\n  mount NewspaperWorks::Engine => '/'\n"
+      end
+    end
+
+    def add_solr_doc
+      inject_into_file 'app/models/solr_document.rb',
+                       after: "include Hyrax::SolrDocumentBehavior" do
+        "\n  include NewspaperWorks::Solr::Document\n"
+      end
+    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 iiif_configuration
+      generate 'newspaper_works:blacklight_iiif_search'
+    end
+
+    # NOTE: BlacklightAdvancedSearch generator installs a view partial by default,
+    # remove it after install, unless app has already customized that view partial
+    def verify_blacklight_adv_search_installed
+      return if IO.read('app/controllers/catalog_controller.rb').include?('include BlacklightAdvancedSearch::Controller')
+      say_status('info', 'INSTALLING BLACKLIGHT ADVANCED SEARCH', :blue)
+      search_form_path = 'app/views/catalog/_search_form.html.erb'
+      existing_search_form = File.exist?(search_form_path) ? true : false
+      generate 'blacklight_advanced_search:install', '--force'
+      remove_file search_form_path unless existing_search_form
+    end
+
+    def advanced_search_configuration
+      generate 'newspaper_works:blacklight_advanced_search'
+    end
+
+    def inject_authorities
+      inject_into_file 'config/authorities/resource_types.yml',
+                       after: "term: Masters Thesis\n" do
+        "  - id: Microfilm\n" \
+        "    term: Microfilm\n" \
+        "  - id: Newspaper\n" \
+        "    term: Newspaper\n"
+      end
+      copy_file "config/authorities/newspaper_article_genres.yml"
+    end
+
+    def catalog_controller_configuration
+      generate 'newspaper_works:catalog_controller'
+    end
+
+    def inject_configuration
+      copy_file 'config/initializers/newspaper_works.rb'
+    end
+
+    def add_helper
+      copy_file "newspaper_works_helper.rb", "app/helpers/newspaper_works_helper.rb"
+    end
+
+    def inject_assets
+      generate 'newspaper_works:assets'
+    end
+  end
+end

data/lib/generators/newspaper_works/templates/annotation_behavior.rb

@@ -0,0 +1,6 @@
+# customizable behavior for IiifSearch
+module BlacklightIiifSearch
+  module AnnotationBehavior
+    include NewspaperWorks::BlacklightIiifSearch::AnnotationBehavior
+  end
+end

data/lib/generators/newspaper_works/templates/config/authorities/newspaper_article_genres.yml

@@ -0,0 +1,86 @@
+# based on hyrax/lib/generators/hyrax/templates/conifg/authorities/resource_types.yml
+terms:
+- id: http://id.loc.gov/vocabulary/graphicMaterials/tgm000098
+  term: "Advertisement"
+  active: true
+- id: http://cv.iptc.org/newscodes/genre/Advice
+  term: "Advice"
+  active: true
+- id: http://cv.iptc.org/newscodes/genre/almanac
+  term: "Almanac"
+  active: true
+- id: http://vocab.getty.edu/aat/300048715
+  term: "Article"
+  active: true
+- id: http://vocab.getty.edu/aat/300265772
+  term: "Birth announcement"
+  active: true
+- id: http://vocab.getty.edu/aat/300191074
+  term: "Caption"
+  active: true
+- id: http://id.loc.gov/vocabulary/graphicMaterials/tgm001739
+  term: "Cartoon (editorial)"
+  active: true
+- id: http://id.loc.gov/vocabulary/graphicMaterials/tgm002334
+  term: "Comics"
+  active: true
+- id: http://vocab.getty.edu/aat/300026441
+  term: "Column"
+  active: true
+- id: http://vocab.getty.edu/aat/300026284
+  term: "Editorial"
+  active: true
+- id: http://cv.iptc.org/newscodes/genre/Feature
+  term: "Feature"
+  active: true
+- id: http://cv.iptc.org/newscodes/genre/Forecast
+  term: "Forecast"
+  active: true
+- id: http://cv.iptc.org/newscodes/genre/horoscope
+  term: "Horoscope"
+  active: true
+- id: http://id.loc.gov/authorities/genreForms/gf2014026110
+  term: "Humor"
+  active: true
+- id: http://id.loc.gov/vocabulary/graphicMaterials/tgm005314
+  term: "Illustration"
+  active: true
+- id: http://vocab.getty.edu/aat/300027757
+  term: "Legal notice"
+  active: true
+- id: http://vocab.getty.edu/aat/300343729
+  term: "Letter to the editor"
+  active: true
+- id: http://id.loc.gov/authorities/genreForms/gf2014026415
+  term: "Literature"
+  active: true
+- id: http://vocab.getty.edu/aat/300311565
+  term: "Masthead"
+  active: true
+- id: http://schema.org/NewsArticle
+  term: "News article"
+  active: true
+- id: http://cv.iptc.org/newscodes/genre/Obituary
+  term: "Obituary/Death announcement"
+  active: true
+- id: http://cv.iptc.org/newscodes/genre/Opinion
+  term: "Opinion"
+  active: true
+- id: http://id.loc.gov/vocabulary/graphicMaterials/tgm007721
+  term: "Photograph"
+  active: true
+- id: http://id.loc.gov/authorities/genreForms/gf2014026158
+  term: "Puzzle/Game"
+  active: true
+- id: http://id.loc.gov/authorities/genreForms/gf2014026168
+  term: "Review"
+  active: true
+- id: http://id.loc.gov/vocabulary/graphicMaterials/tgm009217
+  term: "Schedule/Timetable"
+  active: true
+- id: http://vocab.getty.edu/aat/300136681
+  term: "Shipping record"
+  active: true
+- id: http://vocab.getty.edu/aat/300310119
+  term: "Wedding/Engagement announcement"
+  active: true

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

@@ -0,0 +1,12 @@
+NewspaperWorks.config do |config|
+  # 'publication_unique_id' configs used for Chronicling America style linking
+  #  e.g. First page of an issue from publication with LCCN sn86069873, from Jan. 15, 1897
+  # https://host/newspapers/sn86069873/1897-01-15/ed-1/seq-1
+  # the property that represents a unique identifier for a NewspaperTitle
+  # defaults to :lccn
+  # config.publication_unique_id_property = :lccn
+
+  # the Solr field that represents a unique identifier for a NewspaperTitle
+  # defaults to 'lccn_tesim'
+  # config.publication_unique_id_field = 'lccn_tesim'
+end

data/lib/generators/newspaper_works/templates/config/initializers/patch_blacklight_advanced_search.rb

@@ -0,0 +1,74 @@
+# have to override some methods from BlacklightAdvancedSearch classes/modules
+# to provide date range search functionality
+require BlacklightAdvancedSearch::Engine.root.join(Rails.root, 'config', 'initializers',
+                                                   'patch_blacklight_advanced_search')
+
+class BlacklightAdvancedSearch::QueryParser
+  # override to add date range to query
+  def process_query(params, config)
+    queries = keyword_queries.map do |field, query|
+      ParsingNesting::Tree.parse(query,
+                                 config.advanced_search[:query_parser]).to_query(local_param_hash(field,
+                                                                                                  config))
+    end
+    queries.join(" #{keyword_op} ")
+    return queries if params[:date_start].blank? && params[:date_end].blank?
+    if queries.blank?
+      add_date_range_to_queries(params)
+    else
+      [queries, add_date_range_to_queries(params)].join(' AND ')
+    end
+  end
+
+  # format date input for Solr
+  def add_date_range_to_queries(params)
+    range_start = if params[:date_start].blank? || params[:date_start].match(/[\D]+/)
+                    '*'
+                  else
+                    params[:date_start] + '-01-01T00:00:00.000Z'
+                  end
+    range_end = if params[:date_end].blank? || params[:date_end].match(/[\D]+/)
+                  '*'
+                else
+                  params[:date_end] + '-12-31T23:59:59.999Z'
+                end
+    '(publication_date_dtsi:[' + range_start + ' TO ' + range_end + '])'
+  end
+end
+
+module BlacklightAdvancedSearch::RenderConstraintsOverride
+  # override to add date range to constraints rendering
+  def render_constraints_filters(my_params = params)
+    # these lines are copied from source
+    content = super(my_params)
+    if advanced_query
+      advanced_query.filters.each_pair do |field, value_list|
+        label = facet_field_label(field)
+        content << render_constraint_element(label,
+                                             safe_join(Array(value_list), " <strong class='text-muted constraint-connector'>OR</strong> ".html_safe),
+                                             remove: search_action_path(remove_advanced_filter_group(field, my_params).except(:controller, :action)))
+      end
+      # this is our new line
+      content << render_advanced_date_query(my_params)
+    end
+    content
+  end
+
+  # render the advanced search date query constraints
+  def render_advanced_date_query(localized_params = params)
+    return ''.html_safe if localized_params[:date_start].blank? && localized_params[:date_end].blank?
+    render_constraint_element(t('blacklight.advanced_search.constraints.date'),
+                              date_range_constraints_to_s(localized_params),
+                              classes: ['date_range'],
+                              remove: remove_constraint_url(localized_params.merge(date_start: nil,
+                                                                                   date_end: nil,
+                                                                                   action: 'index')))
+  end
+
+  # render date range constraint text from Advanced Search form
+  def date_range_constraints_to_s(params)
+    return "#{params[:date_end]} or before" if params[:date_start].blank?
+    return "#{params[:date_start]} or later" if params[:date_end].blank?
+    "#{params[:date_start]}-#{params[:date_end]}"
+  end
+end

data/lib/generators/newspaper_works/templates/custom_search_builder.rb

@@ -0,0 +1,23 @@
+# custom SearchBuilder generated by NewspaperWorks; adds behavior to Hyrax::CatalogSearchBuilder:
+# - BlacklightAdvancedSearch::AdvancedSearchBuilder, to support /newspapers_search
+# - NewspaperWorks::HighlightSearchParams, to support highlighting and snippets in results
+# - NewspaperWorks::ExcludeModels, to remove NewspaperTitle, NewspaperContainer,
+#     and NewspaperIssue objects from keyword searches
+class CustomSearchBuilder < Hyrax::CatalogSearchBuilder
+  include BlacklightAdvancedSearch::AdvancedSearchBuilder
+  include NewspaperWorks::HighlightSearchParams
+  include NewspaperWorks::ExcludeModels
+
+  # :exclude_models and :highlight_search_params must be added after advanced_search
+  #   so keyword query input can be properly eval'd
+  self.default_processor_chain += [:add_advanced_parse_q_to_solr, :add_advanced_search_to_solr,
+                                   :exclude_models, :highlight_search_params]
+
+  # add logic to BlacklightAdvancedSearch::AdvancedSearchBuilder
+  # so that date range params are recognized as advanced search
+  # rubocop:disable Naming/PredicateName
+  def is_advanced_search?
+    blacklight_params[:date_start].present? || blacklight_params[:date_end].present? || super
+  end
+  # rubocop:enable Naming/PredicateName
+end

data/lib/generators/newspaper_works/templates/newspaper_works.scss

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

data/lib/generators/newspaper_works/templates/newspaper_works_helper.rb

@@ -0,0 +1,3 @@
+module NewspaperWorksHelper
+  include NewspaperWorks::NewspaperWorksHelperBehavior
+end

data/lib/generators/newspaper_works/templates/search_behavior.rb

@@ -0,0 +1,6 @@
+# customizable behavior for IiifSearch
+module BlacklightIiifSearch
+  module SearchBehavior
+    include NewspaperWorks::BlacklightIiifSearch::SearchBehavior
+  end
+end

data/lib/newspaper_works/configuration.rb

@@ -0,0 +1,14 @@
+module NewspaperWorks
+  class Configuration
+    # 'publication_unique_id' configs used for Chronicling America style linking
+    attr_writer :publication_unique_id_property
+    def publication_unique_id_property
+      @publication_unique_id_property || :lccn
+    end
+
+    attr_writer :publication_unique_id_field
+    def publication_unique_id_field
+      @publication_unique_id_field || 'lccn_tesim'
+    end
+  end
+end

data/lib/newspaper_works/data/fileset_helper.rb

@@ -0,0 +1,25 @@
+module NewspaperWorks
+  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.class == 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.class == String
+        # if "work" context is a FileSet, not actual work, return it
+        return @work if @work.class == FileSet
+        # in most cases, get from work's members:
+        filesets = @work.members.select { |m| m.class == FileSet }
+        filesets.empty? ? nil : filesets[0]
+      end
+    end
+  end
+end

data/lib/newspaper_works/data/path_helper.rb

@@ -0,0 +1,40 @@
+require 'uri'
+
+module NewspaperWorks
+  module Data
+    # Mixin for methods related to paths on filesystem
+    module PathHelper
+      def normalize_path(path)
+        path = path.to_s
+        isuri?(path) ? path : File.expand_path(path)
+      end
+
+      def isuri?(path)
+        !path.scan(URI.regexp).empty?
+      end
+
+      def path_to_uri(path)
+        isuri?(path) ? path : "file://#{path}"
+      end
+
+      def whitelisted_path(path)
+        Hyrax.config.whitelisted_ingest_dirs.any? do |dir|
+          path.start_with?(dir) && path.length > dir.length
+        end
+      end
+
+      def validate_path(path)
+        # treat file URIs equivalent to local paths
+        path = File.expand_path(path.sub(/^file:\/\//, ''))
+        # make sure file exists
+        raise IOError, "Not found: #{path}" unless File.exist?(path)
+        return if whitelisted_path(path)
+        # we cannot use path if it is not whitelisted for Hyrax ingest, we
+        #   would prefer to fail early vs. later+silently
+        raise SecurityError,
+              "Path specified is not configured in Hyrax ingest whitelist: " \
+              "#{path}"
+      end
+    end
+  end
+end