cocina-models 0.75.0 → 0.76.0

data/lib/cocina/models/mapping/normalizers/mods_normalizer.rb

@@ -0,0 +1,409 @@
+# frozen_string_literal: true
+
+module Cocina
+  module Models
+    module Mapping
+      module Normalizers
+        # Normalizes a Fedora MODS document, accounting for differences between Fedora MODS and MODS generated from Cocina.
+        # these adjustments have been approved by our metadata authority, Arcadia.
+        class ModsNormalizer # rubocop:disable Metrics/ClassLength
+          include Cocina::Models::Mapping::Normalizers::Base
+
+          MODS_NS = Cocina::Models::Mapping::FromMods::Description::DESC_METADATA_NS
+          XLINK_NS = Cocina::Models::Mapping::FromMods::Description::XLINK_NS
+
+          # @param [Nokogiri::Document] mods_ng_xml MODS to be normalized
+          # @param [String] druid
+          # @param [String] label
+          # @return [Nokogiri::Document] normalized MODS
+          def self.normalize(mods_ng_xml:, druid:, label:)
+            new(mods_ng_xml: mods_ng_xml, druid: druid, label: label).normalize
+          end
+
+          # @param [Nokogiri::Document] mods_ng_xml MODS to be normalized
+          # @param [String] druid
+          # @return [Nokogiri::Document] normalized MODS
+          def self.normalize_purl(mods_ng_xml:, druid:)
+            new(mods_ng_xml: mods_ng_xml, druid: druid).normalize_purl
+          end
+
+          # @param [Nokogiri::Document] mods_ng_xml MODS to be normalized
+          # @param [String] druid
+          # @param [String] label
+          # @return [Nokogiri::Document] normalized MODS
+          def self.normalize_purl_and_missing_title(mods_ng_xml:, druid:, label:)
+            new(mods_ng_xml: mods_ng_xml, druid: druid, label: label).normalize_purl_and_missing_title
+          end
+
+          # @param [Nokogiri::Document] mods_ng_xml MODS to be normalized
+          # @return [Nokogiri::Document] normalized MODS
+          def self.normalize_identifier_type(mods_ng_xml:)
+            new(mods_ng_xml: mods_ng_xml, druid: nil).normalize_identifier_type
+          end
+
+          def initialize(mods_ng_xml:, druid:, label: nil)
+            @ng_xml = mods_ng_xml.root ? mods_ng_xml.dup : blank_ng_xml
+            @ng_xml.encoding = 'UTF-8'
+            @druid = druid
+            @label = label
+          end
+
+          def normalize
+            normalize_default_namespace
+            normalize_xsi
+            normalize_version
+            normalize_empty_attributes
+            normalize_authority_uris # must be called before OriginInfoNormalizer
+            @ng_xml = Cocina::Models::Mapping::Normalizers::Mods::OriginInfoNormalizer.normalize(mods_ng_xml: ng_xml)
+            @ng_xml = Cocina::Models::Mapping::Normalizers::Mods::SubjectNormalizer.normalize(mods_ng_xml: ng_xml)
+            @ng_xml = Cocina::Models::Mapping::Normalizers::Mods::NameNormalizer.normalize(mods_ng_xml: ng_xml)
+            normalize_related_item_other_type
+            normalize_unmatched_altrepgroup
+            normalize_unmatched_nametitlegroup
+            normalize_xml_space
+            normalize_language_term_type
+            normalize_access_condition
+            normalize_identifier_type_attr
+            normalize_location_physical_location
+            normalize_purl_location
+            normalize_empty_notes
+            @ng_xml = Cocina::Models::Mapping::Normalizers::Mods::TitleNormalizer.normalize(mods_ng_xml: ng_xml, label: label)
+            @ng_xml = Cocina::Models::Mapping::Normalizers::Mods::GeoExtensionNormalizer.normalize(mods_ng_xml: ng_xml, druid: druid)
+            normalize_empty_type_of_resource # Must be after normalize_empty_attributes
+            normalize_notes
+            normalize_abstracts
+            normalize_usage_primary
+            normalize_related_item_attributes
+            # This should be last-ish.
+            normalize_empty_related_items
+            remove_empty_elements(ng_xml.root) # this must be last
+            ng_xml
+          end
+
+          def normalize_purl
+            normalize_purl_location
+            ng_xml
+          end
+
+          def normalize_purl_and_missing_title
+            normalize_purl_location
+            @ng_xml = Cocina::Models::Mapping::Normalizers::Mods::TitleNormalizer.normalize_missing_title(mods_ng_xml: ng_xml, label: label)
+            ng_xml
+          end
+
+          def normalize_identifier_type
+            normalize_identifier_type_attr
+            ng_xml
+          end
+
+          private
+
+          attr_reader :ng_xml, :druid, :label
+
+          # remove all empty elements that have no attributes and no children, recursively
+          def remove_empty_elements(start_node)
+            return unless start_node
+
+            # remove node if there are no element children, there is no text value and there are no attributes
+            if start_node.elements.size.zero? &&
+               start_node.text.blank? &&
+               start_node.attributes.size.zero? &&
+               start_node.name != 'etal'
+              parent = start_node.parent
+              start_node.remove
+              remove_empty_elements(parent) # need to call again after child has been deleted
+            else
+              start_node.element_children.each { |e| remove_empty_elements(e) }
+            end
+          end
+
+          def normalize_default_namespace
+            xml = ng_xml.to_s
+
+            unless xml.include?('xmlns="http://www.loc.gov/mods/v3"')
+              xml.sub!('mods:mods', 'mods:mods xmlns="http://www.loc.gov/mods/v3"')
+              xml.gsub!('mods:', '')
+            end
+
+            regenerate_ng_xml(xml)
+          end
+
+          def normalize_xsi
+            return if ng_xml.namespaces.include?('xmlns:xsi')
+
+            xml = ng_xml.to_s
+            xml.sub!('<mods ', '<mods xmlns:xsi="http://www.w3.org/2001/XMLSchema-instance" ')
+
+            regenerate_ng_xml(xml)
+          end
+
+          def normalize_version
+            # Only normalize version when version isn't mapped.
+            match = /MODS version (\d\.\d)/.match(ng_xml.at('//mods:recordInfo/mods:recordOrigin', mods: MODS_NS)&.content)
+
+            if !match
+              ng_xml.root['version'] = '3.7'
+              ng_xml.root['xsi:schemaLocation'] = 'http://www.loc.gov/mods/v3 http://www.loc.gov/standards/mods/v3/mods-3-7.xsd'
+            elsif match && match[1] != ng_xml.root['version']
+              ng_xml.root['version'] = match[1]
+              ng_xml.root['xsi:schemaLocation'] = "http://www.loc.gov/mods/v3 http://www.loc.gov/standards/mods/v3/mods-#{match[1].sub('.', '-')}.xsd"
+            end
+          end
+
+          def normalize_authority_uris
+            Cocina::Models::Mapping::FromMods::Authority::NORMALIZE_AUTHORITY_URIS.each do |authority_uri|
+              ng_xml.xpath("//mods:*[@authorityURI='#{authority_uri}']", mods: MODS_NS).each do |node|
+                node[:authorityURI] = "#{authority_uri}/"
+              end
+            end
+          end
+
+          def normalize_purl_location
+            normalize_purl_for(ng_xml.root, purl: Cocina::Models::Mapping::Purl.for(druid: druid))
+            ng_xml.xpath('/mods:mods/mods:relatedItem', mods: MODS_NS).each { |related_item_node| normalize_purl_for(related_item_node) }
+          end
+
+          def normalize_purl_for(base_node, purl: nil)
+            purl_nodes(base_node).each do |purl_node|
+              purl_node.content = Cocina::Models::Mapping::FromMods::Purl.purl_value(purl_node)
+            end
+
+            # If there is a purl, add a url node if there is not already one.
+            if purl && purl_nodes(base_node).to_a.none? { |purl_node| purl_node.content == purl }
+              new_location = Nokogiri::XML::Node.new('location', Nokogiri::XML(nil))
+              new_url = Nokogiri::XML::Node.new('url', Nokogiri::XML(nil))
+              new_url.content = purl
+              new_location << new_url
+              base_node << new_location
+            end
+
+            purl_nodes(base_node).each do |purl_node|
+              next if purl_node == Cocina::Models::Mapping::FromMods::Purl.primary_purl_node(base_node, purl)
+
+              # Move into own relatedItem
+              new_related_item = Nokogiri::XML::Node.new('relatedItem', Nokogiri::XML(nil))
+              location_node = purl_node.parent
+              location_node.remove
+              new_related_item << location_node
+              base_node << new_related_item
+              purl_node[:usage] = 'primary display'
+            end
+
+            primary_url_node = primary_url_node_for(base_node, purl)
+            base_node.xpath('mods:location/mods:url', mods: MODS_NS).each do |url_node|
+              if url_node == primary_url_node
+                url_node[:usage] = 'primary display'
+              elsif url_node[:usage] == 'primary display'
+                url_node.delete('usage')
+              end
+            end
+          end
+
+          def purl_nodes(base_node)
+            base_node.xpath('mods:location/mods:url', mods: MODS_NS).select { |url_node| ::Cocina::Models::Mapping::Purl.purl?(url_node.text) }
+          end
+
+          def primary_url_node_for(base_node, purl)
+            primary_purl_nodes, primary_url_nodes = base_node.xpath('mods:location/mods:url[@usage="primary display"]', mods: MODS_NS)
+                                                             .partition { |url_node| ::Cocina::Models::Mapping::Purl.purl?(url_node.text) }
+            all_purl_nodes = base_node.xpath('mods:location/mods:url', mods: MODS_NS)
+                                      .select { |url_node| ::Cocina::Models::Mapping::Purl.purl?(url_node.text) }
+
+            this_purl_node = purl ? all_purl_nodes.find { |purl_node| purl_node.content == purl } : nil
+
+            primary_purl_nodes.first || primary_url_nodes.first || this_purl_node || all_purl_nodes.first
+          end
+
+          def normalize_related_item_other_type
+            ng_xml.xpath('//mods:relatedItem[@type and @otherType]', mods: MODS_NS).each do |related_node|
+              related_node.delete('otherType')
+              related_node.delete('otherTypeURI')
+              related_node.delete('otherTypeAuth')
+            end
+          end
+
+          def normalize_empty_notes
+            ng_xml.xpath('//mods:note[not(text()) and not(@xlink:href)]', mods: MODS_NS, xlink: XLINK_NS).each(&:remove)
+          end
+
+          def normalize_empty_type_of_resource
+            ng_xml.xpath('//mods:typeOfResource[not(text())][not(@*)]', mods: MODS_NS).each(&:remove)
+          end
+
+          def normalize_unmatched_altrepgroup
+            normalize_unmatched_altrepgroup_for(ng_xml.root)
+            ng_xml.xpath('//mods:relatedItem', mods: MODS_NS).each { |related_item_node| normalize_unmatched_altrepgroup_for(related_item_node) }
+          end
+
+          def normalize_unmatched_altrepgroup_for(base_node)
+            ids = {}
+            base_node.xpath('mods:*[@altRepGroup]', mods: MODS_NS).each do |node|
+              id = [node['altRepGroup'], node.name]
+              ids[id] ||= []
+              ids[id] << node
+            end
+
+            ids.each_value do |nodes|
+              next unless nodes.size == 1
+
+              nodes.first.delete('altRepGroup')
+            end
+          end
+
+          def normalize_unmatched_nametitlegroup
+            normalize_unmatched_nametitlegroup_for(ng_xml.root)
+            ng_xml.xpath('//mods:relatedItem', mods: MODS_NS).each { |related_item_node| normalize_unmatched_nametitlegroup_for(related_item_node) }
+          end
+
+          def normalize_unmatched_nametitlegroup_for(base_node)
+            ids = {}
+            base_node.xpath('mods:name[@nameTitleGroup] | mods:titleInfo[@nameTitleGroup]', mods: MODS_NS).each do |node|
+              id = node['nameTitleGroup']
+              ids[id] ||= []
+              ids[id] << node
+            end
+
+            ids.each_value do |nodes|
+              next unless nodes.size == 1
+
+              nodes.first.delete('nameTitleGroup')
+            end
+          end
+
+          def normalize_empty_attributes
+            ng_xml.xpath('//mods:*[@*=""]', mods: MODS_NS).each do |node|
+              node.each { |attr_name, attr_value| node.delete(attr_name) if attr_value.blank? }
+            end
+          end
+
+          def normalize_xml_space
+            ng_xml.xpath('//mods:*[@xml:space]', mods: MODS_NS).each do |node|
+              node.delete('space')
+            end
+          end
+
+          def normalize_language_term_type
+            ng_xml.xpath('//mods:languageTerm[not(@type)]', mods: MODS_NS).each do |node|
+              node['type'] = 'code'
+            end
+          end
+
+          def normalize_access_condition
+            ng_xml.xpath('//mods:accessCondition[@type="restrictionOnAccess"]', mods: MODS_NS).each do |node|
+              node['type'] = 'restriction on access'
+            end
+            ng_xml.xpath('//mods:accessCondition[@type="restrictionsOnAccess"]', mods: MODS_NS).each do |node|
+              node['type'] = 'restriction on access'
+            end
+            ng_xml.xpath('//mods:accessCondition[@type="useAndReproduction"]', mods: MODS_NS).each do |node|
+              node['type'] = 'use and reproduction'
+            end
+          end
+
+          def normalize_identifier_type_attr
+            ng_xml.xpath('//mods:identifier[@type]', mods: MODS_NS).each do |node|
+              node['type'] = normalized_identifier_type_for(node['type'])
+            end
+            ng_xml.xpath('//mods:nameIdentifier[@type]', mods: MODS_NS).each do |node|
+              node['type'] = normalized_identifier_type_for(node['type'])
+            end
+            ng_xml.xpath('//mods:recordIdentifier[@source]', mods: MODS_NS).each do |node|
+              node['source'] = normalized_identifier_type_for(node['source'])
+            end
+          end
+
+          def normalized_identifier_type_for(type)
+            cocina_type, _mods_type, identifier_source = Cocina::Models::Mapping::FromMods::IdentifierType.cocina_type_for_mods_type(type)
+
+            return Cocina::Models::Mapping::FromMods::IdentifierType.mods_type_for_cocina_type(cocina_type) if identifier_source
+
+            type
+          end
+
+          def normalize_location_physical_location
+            ng_xml.xpath('//mods:location', mods: MODS_NS).each do |location_node|
+              location_node.xpath('mods:physicalLocation|mods:url|mods:shelfLocator', mods: MODS_NS).each do |node|
+                next unless node.content.present? || node['xlink:href']
+
+                new_location = Nokogiri::XML::Node.new('location', ng_xml)
+                new_location << node
+                location_node.parent << new_location
+              end
+              location_node.remove
+            end
+          end
+
+          def normalize_empty_related_items
+            ng_xml.xpath('//mods:relatedItem/mods:part[count(mods:*)=1]/mods:detail[count(mods:*)=1]/mods:number[not(text())]',
+                         mods: MODS_NS).each do |number_node|
+              number_node.parent.parent.remove
+            end
+            ng_xml.xpath('//mods:relatedItem[not(mods:*) and not(@xlink:href)]', mods: MODS_NS, xlink: XLINK_NS).each(&:remove)
+          end
+
+          def normalize_notes
+            ng_xml.xpath('//mods:note', mods: MODS_NS).each do |note_node|
+              if Cocina::Models::Mapping::ToMods::Note.note_type_to_abstract_type.include?(note_node['type']&.downcase) ||
+                 Cocina::Models::Mapping::ToMods::Note.display_label_to_abstract_type.include?(note_node['displayLabel'])
+                note_node.delete('type') unless note_node['type']&.downcase == 'summary'
+                note_node.name = 'abstract'
+              end
+              if Cocina::Models::Mapping::ToMods::Note.display_label_to_abstract_type.include? note_node['displayLabel']
+                note_node['displayLabel'] =
+                  note_node['displayLabel'].capitalize
+              end
+            end
+          end
+
+          def normalize_abstracts
+            ng_xml.xpath('/mods:mods/mods:abstract', mods: MODS_NS).each do |abstract_node|
+              if Cocina::Models::Mapping::ToMods::Note.note_type_to_abstract_type.include? abstract_node['type']&.downcase
+                abstract_node['type'] =
+                  abstract_node['type'].downcase
+              end
+              if Cocina::Models::Mapping::ToMods::Note.display_label_to_abstract_type.include? abstract_node['displayLabel']
+                abstract_node['displayLabel'] =
+                  abstract_node['displayLabel'].capitalize
+              end
+              abstract_node.delete('type') if abstract_node['type'] == 'abstract'
+            end
+          end
+
+          def normalize_usage_primary
+            normalize_usage_primary_for(ng_xml.root)
+            ng_xml.xpath('/mods:mods/mods:relatedItem', mods: ModsNormalizer::MODS_NS).each do |related_item_node|
+              normalize_usage_primary_for(related_item_node)
+            end
+            ng_xml.xpath('//mods:subject', mods: ModsNormalizer::MODS_NS).each { |subject_node| normalize_usage_primary_for(subject_node) }
+          end
+
+          def normalize_usage_primary_for(base_node)
+            %w[genre language classification subject titleInfo typeOfResource name].each do |node_name|
+              primary_nodes = base_node.xpath("mods:#{node_name}[@usage=\"primary\"]", mods: MODS_NS)
+              next if primary_nodes.size < 2
+
+              primary_nodes[1..].each { |primary_node| primary_node.delete('usage') }
+            end
+          end
+
+          def normalize_related_item_attributes
+            ng_xml.xpath('/mods:mods/mods:relatedItem[@lang or @script]', mods: ModsNormalizer::MODS_NS).each do |related_item_node|
+              related_item_node.delete('lang')
+              related_item_node.delete('script')
+            end
+          end
+
+          def blank_ng_xml
+            Nokogiri::XML(<<~XML
+              <mods xmlns="http://www.loc.gov/mods/v3"#{' '}
+                xmlns:xsi="http://www.w3.org/2001/XMLSchema-instance"#{' '}
+                version="3.6"#{' '}
+                xsi:schemaLocation="http://www.loc.gov/mods/v3 http://www.loc.gov/standards/mods/v3/mods-3-6.xsd" />
+            XML
+                         )
+          end
+        end
+      end
+    end
+  end
+end

data/lib/cocina/models/mapping/purl.rb

@@ -0,0 +1,28 @@
+# frozen_string_literal: true
+
+module Cocina
+  module Models
+    module Mapping
+      # Utility methods for generating purl links
+      class Purl
+        class_attribute :base_url
+        self.base_url = 'https://purl.stanford.edu'
+
+        def self.for(druid:)
+          return nil if druid.nil?
+
+          "#{base_url}/#{druid.delete_prefix('druid:')}"
+        end
+
+        def self.purl?(node)
+          node.start_with?("https://#{host}") || node.start_with?("http://#{host}")
+        end
+
+        # the purl without the protocol part
+        def self.host
+          @host ||= base_url.sub(%r{^https?://}, '')
+        end
+      end
+    end
+  end
+end

data/lib/cocina/models/mapping/to_mods/access.rb

@@ -0,0 +1,155 @@
+# frozen_string_literal: true
+
+module Cocina
+  module Models
+    module Mapping
+      module ToMods
+        # Maps the Access subschema from cocina to MODS XML
+        class Access # rubocop:disable Metrics/ClassLength
+          # @params [Nokogiri::XML::Builder] xml
+          # @params [Cocina::Models::Access] access
+          # @params [string] purl
+          def self.write(xml:, access:, purl:)
+            new(xml: xml, access: access, purl: purl).write
+          end
+
+          def initialize(xml:, access:, purl:)
+            @xml = xml
+            @access = access
+            @purl = purl
+          end
+
+          def write
+            write_purl unless purl.nil?
+            return if access.nil?
+
+            write_access_conditions if access
+
+            Array(access.url).each do |url|
+              xml.location do
+                write_url(url)
+              end
+            end
+
+            write_physical_locations
+            write_digital_locations
+            write_shelf_locators
+            write_access_contact_locations
+          end
+
+          private
+
+          attr_reader :xml, :access, :purl
+
+          def write_physical_locations
+            Array(access.physicalLocation).reject do |physical_location|
+              shelf_locator?(physical_location)
+            end.each do |physical_location|
+              xml.location do
+                xml.physicalLocation physical_location.value || physical_location.code,
+                                     descriptive_attrs(physical_location)
+              end
+            end
+          end
+
+          def write_digital_locations
+            Array(access.digitalLocation).select do |digital_location|
+              digital_location.type == 'discovery'
+            end.each do |digital_location|
+              xml.location do
+                xml.physicalLocation digital_location.value || digital_location.code,
+                                     descriptive_attrs(digital_location)
+              end
+            end
+          end
+
+          def write_access_contact_locations
+            Array(access.accessContact).each do |access_contact|
+              if access_contact.type == 'email'
+                xml.note access_contact.value, descriptive_attrs(access_contact).merge({ type: 'contact' })
+              else
+                xml.location do
+                  xml.physicalLocation access_contact.value || access_contact.code,
+                                       { type: 'repository' }.merge(descriptive_attrs(access_contact))
+                end
+              end
+            end
+          end
+
+          def write_shelf_locators
+            Array(access.physicalLocation).select do |physical_location|
+              shelf_locator?(physical_location)
+            end.each do |physical_location|
+              xml.location do
+                xml.shelfLocator physical_location.value
+              end
+            end
+          end
+
+          def write_url(url)
+            url_attrs = {}.tap do |attrs|
+              attrs[:usage] = 'primary display' if url.status == 'primary'
+              attrs[:displayLabel] = url.displayLabel
+              attrs[:note] = url.note.first.value if url.note.present?
+            end.compact
+            xml.url url.value, url_attrs
+          end
+
+          def primary_url_is_not_purl?
+            Array(access&.url).any? { |url| url.status == 'primary' }
+          end
+
+          def write_purl
+            purl_attrs = {}.tap do |attrs|
+              attrs[:note] = find_note_value(nil)
+              attrs[:usage] = 'primary display' unless primary_url_is_not_purl?
+              attrs[:displayLabel] = find_note_value('display label')
+            end.compact
+
+            xml.location do
+              xml.url purl, purl_attrs
+            end
+          end
+
+          def find_note_value(note_type)
+            Array(access&.note).find do |note|
+              note.type == note_type && purl_note?(note)
+            end&.value
+          end
+
+          def descriptive_attrs(cocina)
+            {
+              valueURI: cocina.uri,
+              authorityURI: cocina.source&.uri,
+              authority: cocina.source&.code,
+              script: cocina.valueLanguage&.valueScript&.code,
+              lang: cocina.valueLanguage&.code,
+              type: cocina.type,
+              displayLabel: cocina.displayLabel,
+              'xlink:href' => cocina.valueAt
+            }.compact
+          end
+
+          def shelf_locator?(physical_location)
+            physical_location.type == 'shelf locator'
+          end
+
+          def write_access_conditions
+            Array(access.note).reject { |note| purl_note?(note) }.each do |note|
+              attributes = {
+                type: note.type == 'access restriction' ? 'restriction on access' : note.type,
+                displayLabel: note.displayLabel,
+                'xlink:href' => note.valueAt
+              }.compact
+              xml.accessCondition note.value, attributes
+            end
+          end
+
+          def purl_note?(note)
+            Array(note.appliesTo).any? { |applies_to| applies_to.value == 'purl' }
+          end
+        end
+      end
+    end
+  end
+end