geo_concerns 0.0.4 → 0.0.5

data/app/services/geo_concerns/discovery/document_builder/basic_metadata_builder.rb

@@ -0,0 +1,69 @@
+module GeoConcerns
+  module Discovery
+    class DocumentBuilder
+      class BasicMetadataBuilder
+        attr_reader :geo_concern
+
+        def initialize(geo_concern)
+          @geo_concern = geo_concern
+        end
+
+        # Builds fields such as id, subject, and publisher.
+        # @param [AbstractDocument] discovery document
+        def build(document)
+          build_simple_attributes(document)
+          build_complex_attributes(document)
+        end
+
+        private
+
+          # Returns an array of attributes to add to document.
+          # @return [Array] attributes
+          def simple_attributes
+            [:id, :creator, :subject, :spatial, :temporal,
+             :title, :provenance, :language, :publisher]
+          end
+
+          # Builds simple metadata attributes.
+          # @param [AbstractDocument] discovery document
+          def build_simple_attributes(document)
+            simple_attributes.each do |a|
+              document.send("#{a}=", geo_concern.send(a.to_s))
+            end
+          end
+
+          # Builds more complex metadata attributes.
+          # @param [AbstractDocument] discovery document
+          def build_complex_attributes(document)
+            document.identifier = identifier
+            document.description = description
+            document.access_rights = geo_concern.solr_document.visibility
+            document.slug = slug
+          end
+
+          # Returns the work description. If none is available,
+          # a basic description is created.
+          # @return [String] description
+          def description
+            return geo_concern.description.first if geo_concern.description.first
+            "A #{geo_concern.human_readable_type.downcase} object."
+          end
+
+          # Returns the document slug for use in discovery systems.
+          # @return [String] document slug
+          def slug
+            return unless geo_concern.provenance
+            "#{geo_concern.provenance[0].downcase.tr(' ', '-')}-#{geo_concern.id}"
+          end
+
+          # Returns the document identifier. If there is no identifier,
+          # the work id is used instead.
+          # @return [String] document identifier
+          def identifier
+            field = geo_concern.solr_document[:identifier_tesim]
+            field ? field.first : geo_concern.id
+          end
+      end
+    end
+  end
+end

data/app/services/geo_concerns/discovery/document_builder/composite_builder.rb

@@ -0,0 +1,21 @@
+module GeoConcerns
+  module Discovery
+    class DocumentBuilder
+      class CompositeBuilder
+        attr_reader :services
+
+        def initialize(*services)
+          @services = services.compact
+        end
+
+        # Runs each builder service to build a discovery document.
+        # @param [AbstractDocument] discovery document
+        def build(document)
+          services.each do |service|
+            service.build(document)
+          end
+        end
+      end
+    end
+  end
+end

data/app/services/geo_concerns/discovery/document_builder/date_builder.rb

@@ -0,0 +1,38 @@
+module GeoConcerns
+  module Discovery
+    class DocumentBuilder
+      class DateBuilder
+        attr_reader :geo_concern
+
+        def initialize(geo_concern)
+          @geo_concern = geo_concern
+        end
+
+        # Builds date fields such as layer year and modified date.
+        # @param [AbstractDocument] discovery document
+        def build(document)
+          document.layer_year = layer_year
+          document.date_modified = date_modified
+        end
+
+        private
+
+          # Returns a year associated with the layer. Taken from first
+          # value in temporal or from date uploaded. If neither is valid,
+          # the current year is used.
+          # @return [Integer] year
+          def layer_year
+            date = geo_concern.temporal.first || geo_concern.date_uploaded
+            year = date.match(/(?<=\D|^)(\d{4})(?=\D|$)/)
+            year ? year[0].to_i : Time.current.year
+          end
+
+          # Returns the date the work was modified.
+          # @return [String] date in rfc3339 format.
+          def date_modified
+            DateTime.rfc3339(geo_concern.solr_document[:date_modified_dtsi]).to_s
+          end
+      end
+    end
+  end
+end

data/app/services/geo_concerns/discovery/document_builder/document_helper.rb

@@ -0,0 +1,10 @@
+module GeoConcerns
+  module Discovery
+    class DocumentBuilder
+      class DocumentHelper
+        include Rails.application.routes.url_helpers
+        include ActionDispatch::Routing::PolymorphicRoutes
+      end
+    end
+  end
+end

data/app/services/geo_concerns/discovery/document_builder/document_path.rb

@@ -0,0 +1,82 @@
+module GeoConcerns
+  module Discovery
+    class DocumentBuilder
+      class DocumentPath
+        attr_reader :geo_concern, :ssl
+        def initialize(geo_concern, ssl: false)
+          @geo_concern = geo_concern
+          @ssl = ssl
+        end
+
+        # Returns url for geo concern show page.
+        # @return [String] geo concern show page url
+        def to_s
+          helper.polymorphic_url(geo_concern, protocol: protocol)
+        end
+
+        # Returns url for downloading the original file.
+        # @return [String] original file download url
+        def file_download
+          return unless file_set
+          helper.download_url(file_set, protocol: protocol)
+        end
+
+        # Returns url for downloading the metadata file.
+        # @param [String] metadata file format to download
+        # @return [String] metadata download url
+        def metadata_download(format)
+          return unless metadata_file_set
+          path = helper.download_url(metadata_file_set, protocol: protocol)
+          mime_type = metadata_file_set.solr_document[:geo_mime_type_ssim].first
+          path if MetadataFormatService.label(mime_type) == format
+        end
+
+        # Returns url for thumbnail image.
+        # @return [String] thumbnail url
+        def thumbnail
+          return unless file_download
+          "#{file_download}?file=thumbnail"
+        end
+
+        private
+
+          # Returns the first geo file set presenter attached to work.
+          # @return [FileSetPresenter] geo file set presenter
+          def file_set
+            return unless geo_concern.geo_file_set_presenters
+            geo_concern.geo_file_set_presenters.first
+          end
+
+          # Returns the first metadata file set presenter attached to work.
+          # @return [FileSetPresenter] metadata file set presenter
+          def metadata_file_set
+            return unless geo_concern.external_metadata_file_set_presenters
+            geo_concern.external_metadata_file_set_presenters.first
+          end
+
+          # Instantiates a DocumentHelper object.
+          # Used for access to url_helpers and poymorphic routes.
+          # @return [DocumentHelper] document helper
+          def helper
+            @helper ||= DocumentHelper.new
+          end
+
+          # Indicates if the ssl is enabled.
+          # @return [Boolean] use ssl
+          def ssl?
+            @ssl == true
+          end
+
+          # Returns protocol to use in url. Depends on ssl status.
+          # @return [Boolean] http protocol
+          def protocol
+            if ssl?
+              :https
+            else
+              :http
+            end
+          end
+      end
+    end
+  end
+end

data/app/services/geo_concerns/discovery/document_builder/layer_info_builder.rb

@@ -0,0 +1,65 @@
+module GeoConcerns
+  module Discovery
+    class DocumentBuilder
+      class LayerInfoBuilder
+        attr_reader :geo_concern
+
+        def initialize(geo_concern)
+          @geo_concern = geo_concern
+        end
+
+        # Builds fields about the geospatial file such as geometry and format.
+        # @param [AbstractDocument] discovery document
+        def build(document)
+          document.geom_type = geom_type
+          document.format = format
+        end
+
+        private
+
+          # Uses parent work class to determine file geometry type.
+          # These geom types are used in geoblacklight documents.
+          # @return [String] file geometry type
+          def geom_type
+            case geo_concern.class.to_s
+            when "GeoConcerns::ImageWorkShowPresenter"
+              'Scanned Map'
+            when "GeoConcerns::RasterWorkShowPresenter"
+              'Raster'
+            when "GeoConcerns::VectorWorkShowPresenter"
+              vector_geom_type
+            end
+          end
+
+          # Uses file mime type to determine file format.
+          # @return [String] file format code
+          def format
+            if ImageFormatService.include? geo_mime_type
+              ImageFormatService.code(geo_mime_type)
+            elsif RasterFormatService.include? geo_mime_type
+              RasterFormatService.code(geo_mime_type)
+            elsif VectorFormatService.include? geo_mime_type
+              VectorFormatService.code(geo_mime_type)
+            end
+          end
+
+          # Returns the geometry for a vector file.
+          # @return [String] vector geometry
+          def vector_geom_type
+            # TODO: Get the geom type as part of a geo characterization service.
+            # "Point, Line, Polygon, Mixed"
+            'Mixed'
+          end
+
+          # Returns the 'geo' mime type of the first file attached to the work.
+          # @return [String] file mime type
+          def geo_mime_type
+            file_sets = geo_concern.geo_file_set_presenters
+            return if file_sets.nil? || file_sets.empty?
+            return unless (mime_types = file_sets.first.solr_document[:geo_mime_type_ssim])
+            mime_types.first
+          end
+      end
+    end
+  end
+end

data/app/services/geo_concerns/discovery/document_builder/references_builder.rb

@@ -0,0 +1,82 @@
+module GeoConcerns
+  module Discovery
+    class DocumentBuilder
+      class ReferencesBuilder
+        attr_reader :geo_concern, :path
+
+        def initialize(geo_concern, path)
+          @geo_concern = geo_concern
+          @path = path
+        end
+
+        # Builds service reference fields such as thumbnail and download url.
+        # @param [AbstractDocument] discovery document
+        def build(document)
+          document.wxs_identifier = wxs_identifier
+          build_metadata_refs(document)
+          build_download_refs(document)
+        end
+
+        private
+
+          # Builds metadata file references.
+          # @param [AbstractDocument] discovery document
+          def build_metadata_refs(document)
+            document.fgdc = fgdc
+            document.iso19139 = iso19139
+            document.mods = mods
+          end
+
+          # Builds geospatial file download references.
+          # @param [AbstractDocument] discovery document
+          def build_download_refs(document)
+            document.download = download
+            document.url = url
+            document.thumbnail = thumbnail
+          end
+
+          # Returns the identifier to use with WMS/WFS/WCS services.
+          # @return [String] wxs indentifier
+          def wxs_identifier
+            geo_concern.id
+          end
+
+          # Returns a url to access further descriptive information.
+          # @return [String] work show page url
+          def url
+            path.to_s
+          end
+
+          # Returns a direct file download url
+          # @return [String] direct file  url
+          def download
+            path.file_download
+          end
+
+          # Returns an FGDC metadata document download url
+          # @return [String] FGDC metadata document url
+          def fgdc
+            path.metadata_download('FGDC')
+          end
+
+          # Returns an ISO19139 metadata document download url
+          # @return [String] ISO19139 metadata document url
+          def iso19139
+            path.metadata_download('ISO19139')
+          end
+
+          # Returns an MODS metadata document download url
+          # @return [String] MODS metadata document url
+          def mods
+            path.metadata_download('MODS')
+          end
+
+          # Returns a thumbnail file url
+          # @return [String] thumbnail url
+          def thumbnail
+            path.thumbnail
+          end
+      end
+    end
+  end
+end

data/app/services/geo_concerns/discovery/document_builder/spatial_builder.rb

@@ -0,0 +1,40 @@
+module GeoConcerns
+  module Discovery
+    class DocumentBuilder
+      class SpatialBuilder
+        attr_reader :geo_concern
+
+        def initialize(geo_concern)
+          @geo_concern = geo_concern
+        end
+
+        # Builds spatial fields such as bounding box and solr geometry.
+        # @param [AbstractDocument] discovery document
+        def build(document)
+          document.geo_rss_coverage = to_geo_rss
+          document.solr_coverage = to_solr
+        end
+
+        private
+
+          # Parses coverage field from geo work and instantiates a coverage object.
+          # @return [GeoConcerns::Coverage] coverage object
+          def coverage
+            @coverage ||= GeoConcerns::Coverage.parse(geo_concern.coverage.first)
+          end
+
+          # Returns the coverage as georss.
+          # @return [String] coverage in georss format
+          def to_geo_rss
+            "#{coverage.s} #{coverage.w} #{coverage.n} #{coverage.e}"
+          end
+
+          # Returns the coverage in solr format.
+          # @return [String] coverage in solr format
+          def to_solr
+            "ENVELOPE(#{coverage.w}, #{coverage.e}, #{coverage.n}, #{coverage.s})"
+          end
+      end
+    end
+  end
+end

data/app/services/geo_concerns/discovery/geoblacklight_document.rb

@@ -0,0 +1,122 @@
+require 'json-schema'
+
+module GeoConcerns
+  module Discovery
+    class GeoblacklightDocument < AbstractDocument
+      # Implements the to_hash method on the abstract document.
+      # @param _args [Array<Object>] arguments needed for the renderer, unused here
+      # @return [Hash] geoblacklight document as a hash
+      def to_hash(_args = nil)
+        document
+      end
+
+      # Implements the to_json method on the abstract document.
+      # @param _args [Array<Object>] arguments needed for the json renderer, unused here
+      # @return [String] geoblacklight document as a json string
+      def to_json(_args = nil)
+        document.to_json
+      end
+
+      private
+
+        # Builds the geoblacklight document hash.
+        # @return [Hash] geoblacklight document as a hash
+        # rubocop:disable Metrics/MethodLength, Metrics/AbcSize
+        def document_hash
+          {
+            uuid: id,
+            dc_identifier_s: identifier,
+            dc_title_s: title.first,
+            dc_description_s: description,
+            dc_rights_s: rights,
+            dct_provenance_s: provenance.first,
+            dc_creator_sm: creator,
+            dc_language_s: language.first,
+            dc_publisher_s: publisher.first,
+            dc_subject_sm: subject,
+            dct_spatial_sm: spatial,
+            dct_temporal_sm: temporal,
+            layer_slug_s: slug,
+            georss_box_s: geo_rss_coverage,
+            solr_geom: solr_coverage,
+            solr_year_i: layer_year,
+            layer_modified_dt: date_modified,
+            layer_id_s: wxs_identifier,
+            dct_references_s: clean_document(references).to_json,
+            layer_geom_type_s: geom_type,
+            dc_format_s: process_format_codes(format)
+          }
+        end
+        # rubocop:enable Metrics/LineLength, Metrics/AbcSize
+
+        # Builds the dct_references hash.
+        # @return [Hash] geoblacklight references as a hash
+        def references
+          {
+            'http://schema.org/url' => url,
+            'http://www.opengis.net/cat/csw/csdgm' => fgdc,
+            'http://www.isotc211.org/schemas/2005/gmd/' => iso19139,
+            'http://www.loc.gov/mods/v3' => mods,
+            'http://schema.org/downloadUrl' => download,
+            'http://schema.org/thumbnailUrl' => thumbnail
+          }
+        end
+
+        # Returns the geoblacklight rights field based on work visibility.
+        # @return [String] geoblacklight access rights
+        def rights
+          if access_rights == Hydra::AccessControls::AccessRight::VISIBILITY_TEXT_VALUE_PUBLIC
+            'Public'
+          else
+            'Restricted'
+          end
+        end
+
+        # Transforms shapfile, tiff, and arc grid format codes into geoblacklight format codes.
+        # @return [String] geoblacklight format codes
+        def process_format_codes(format)
+          case format
+          when 'ESRI Shapefile'
+            'Shapefile'
+          when 'GTiff'
+            'GeoTIFF'
+          when 'AIG'
+            'ArcGRID'
+          else
+            format
+          end
+        end
+
+        # Returns the location of geoblacklight json schema document.
+        # @return [String] geoblacklight json schema document path
+        def schema
+          Rails.root.join('config', 'discovery', 'geoblacklight_schema.json').to_s
+        end
+
+        # Validates the geoblacklight document against the json schema.
+        # @return [Boolean] is the document valid?
+        def valid?(doc)
+          JSON::Validator.validate(schema, doc, validate_schema: true)
+        end
+
+        # Returns a hash of errors from json schema validation.
+        # @return [Hash] json schema validation errors
+        def schema_errors(doc)
+          { error: JSON::Validator.fully_validate(schema, doc) }
+        end
+
+        # Cleans the geoblacklight document hash by removing unused fields,
+        # then validates it again a json schema. If there are errors, an
+        # error hash is returned, otherwise, the cleaned doc is returned.
+        # @return [Hash] geoblacklight document hash or error hash
+        def document
+          clean = clean_document(document_hash)
+          if valid?(clean)
+            clean
+          else
+            schema_errors(clean)
+          end
+        end
+    end
+  end
+end