cocina-models 0.74.1 → 0.77.0

data/lib/cocina/models/mapping/normalizers/mods/origin_info_normalizer.rb

@@ -0,0 +1,157 @@
+# frozen_string_literal: true
+
+module Cocina
+  module Models
+    module Mapping
+      module Normalizers
+        module Mods
+          # Normalizes a Fedora MODS document for originInfo elements.
+          # Must be called after authorityURI attribs are normalized
+          class OriginInfoNormalizer
+            DATE_FIELDS = %w[dateIssued copyrightDate dateCreated dateCaptured dateValid dateOther dateModified].freeze
+
+            # @param [Nokogiri::Document] mods_ng_xml MODS to be normalized
+            # @return [Nokogiri::Document] normalized MODS
+            def self.normalize(mods_ng_xml:)
+              new(mods_ng_xml: mods_ng_xml).normalize
+            end
+
+            def initialize(mods_ng_xml:)
+              @ng_xml = mods_ng_xml.dup
+              @ng_xml.encoding = 'UTF-8'
+            end
+
+            def normalize
+              remove_empty_child_elements
+              remove_empty_origin_info # must be after remove_empty_child_elements
+              normalize_legacy_mods_event_type
+              place_term_type_normalization
+              place_term_authority_normalization # must be after place_term_type_normalization
+              normalize_authority_marcountry
+              single_key_date
+              remove_trailing_period_from_date_values
+              ng_xml
+            end
+
+            private
+
+            attr_reader :ng_xml
+
+            # must be called before remove_empty_origin_info
+            def remove_empty_child_elements
+              ng_xml.root.xpath('//mods:originInfo/mods:*', mods: ModsNormalizer::MODS_NS).each do |child_node|
+                # if a node has either of these 2 attributes, it could have meaning even without any content
+                next if child_node.xpath('.//*[@valueURI]').present?
+                next if child_node.xpath('.//*[@xlink:href]', xlink: ModsNormalizer::XLINK_NS).present?
+
+                child_node.remove if child_node.content.blank?
+              end
+            end
+
+            # must be called after remove_empty_child_elements
+            def remove_empty_origin_info
+              ng_xml.root.xpath('//mods:originInfo[not(mods:*) and not(@*)]', mods: ModsNormalizer::MODS_NS).each(&:remove)
+              # make sure we remove ones such as <originInfo eventType="publication"/>
+              ng_xml.root.xpath('//mods:originInfo[not(mods:*) and not(text()[normalize-space()])]', mods: ModsNormalizer::MODS_NS).each(&:remove)
+            end
+
+            LEGACY_EVENT_TYPES_2_TYPE = Cocina::Models::Mapping::FromMods::Event::LEGACY_EVENT_TYPES_2_TYPE
+
+            # because eventType is a relatively new addition to the MODS schema, records converted from MARC to MODS prior
+            #   to its introduction used displayLabel as a stopgap measure, with certain values
+            # The same values were also sometimes used as eventType values themselves, and will be converted to our preferred vocab.
+            def normalize_legacy_mods_event_type
+              ng_xml.root.xpath('//mods:originInfo[@*]', mods: ModsNormalizer::MODS_NS).each do |origin_info_node|
+                event_type = origin_info_node['eventType']
+                event_type = origin_info_node['displayLabel'] if event_type.blank? &&
+                                                                 LEGACY_EVENT_TYPES_2_TYPE.key?(origin_info_node['displayLabel'])
+                event_type = LEGACY_EVENT_TYPES_2_TYPE[event_type] if LEGACY_EVENT_TYPES_2_TYPE.key?(event_type)
+
+                origin_info_node['eventType'] = event_type if event_type.present?
+                origin_info_node.delete('displayLabel') if event_type.present? &&
+                                                           event_type == LEGACY_EVENT_TYPES_2_TYPE[origin_info_node['displayLabel']]
+              end
+            end
+
+            # must be called before place_term_authority_normalization
+            # if the cocina model doesn't have a code, then it will have a value;
+            #   this is output as attribute type=text on the roundtripped placeTerm element
+            def place_term_type_normalization
+              ng_xml.root.xpath('//mods:originInfo/mods:place/mods:placeTerm', mods: ModsNormalizer::MODS_NS).each do |place_term_node|
+                next if place_term_node.content.blank?
+
+                place_term_node['type'] = 'text' if place_term_node.attributes['type'].blank?
+              end
+            end
+
+            # must be called after place_term_type_normalization
+            # if the MODS has a single place element with both text and code placeTerm elements, if the text
+            # element has no authority attributes but the code element DOES have authority attributes, then both
+            # the text and the code elements get the authority attributes from the code element.
+            def place_term_authority_normalization
+              ng_xml.root.xpath('//mods:originInfo/mods:place[mods:placeTerm/@type]', mods: ModsNormalizer::MODS_NS).each do |place_node|
+                text_place_term_node = place_node.xpath("mods:placeTerm[not(@type='code')]", mods: ModsNormalizer::MODS_NS).first
+                next unless text_place_term_node
+                next if text_place_term_node.text.blank?
+
+                code_place_term_node = place_node.xpath("mods:placeTerm[@type='code']", mods: ModsNormalizer::MODS_NS).first
+                next unless code_place_term_node
+                next if code_place_term_node.text.blank?
+
+                text_authority_attributes = authority_attributes(text_place_term_node)
+                code_authority_attributes = authority_attributes(code_place_term_node)
+
+                # NOTE: deliberately skipping situation where text node has some authority info and code node
+                #  has other authority info as we may never encounter this
+
+                if text_authority_attributes.present? && code_authority_attributes.blank?
+                  text_authority_attributes.each do |key, val|
+                    code_place_term_node[key] = val
+                  end
+                  next
+                end
+
+                next if code_authority_attributes.blank? || text_authority_attributes.present?
+
+                code_authority_attributes.each do |key, val|
+                  text_place_term_node[key] = val
+                end
+              end
+            end
+
+            def authority_attributes(ng_node)
+              {
+                valueURI: ng_node['valueURI'],
+                authority: ng_node['authority'],
+                authorityURI: ng_node['authorityURI']
+              }.compact
+            end
+
+            def normalize_authority_marcountry
+              ng_xml.root.xpath("//mods:*[@authority='marcountry']", mods: ModsNormalizer::MODS_NS).each do |node|
+                node[:authority] = 'marccountry'
+              end
+            end
+
+            def single_key_date
+              DATE_FIELDS.each do |date_field|
+                key_date_nodes = ng_xml.root.xpath("//mods:originInfo/mods:#{date_field}[@point and @keyDate='yes']", mods: ModsNormalizer::MODS_NS)
+                next unless key_date_nodes.size == 2
+
+                end_node = key_date_nodes.find { |node| node['point'] == 'end' }
+                end_node.delete('keyDate') if end_node && end_node['keyDate'].present?
+              end
+            end
+
+            def remove_trailing_period_from_date_values
+              DATE_FIELDS.each do |date_field|
+                ng_xml.root.xpath("//mods:originInfo/mods:#{date_field}", mods: ModsNormalizer::MODS_NS)
+                      .each { |date_node| date_node.content = date_node.content.delete_suffix('.') }
+              end
+            end
+          end
+        end
+      end
+    end
+  end
+end

data/lib/cocina/models/mapping/normalizers/mods/subject_normalizer.rb

@@ -0,0 +1,296 @@
+# frozen_string_literal: true
+
+module Cocina
+  module Models
+    module Mapping
+      module Normalizers
+        module Mods
+          # Normalizes a Fedora MODS document for subject elements.
+          class SubjectNormalizer # rubocop:disable Metrics/ClassLength
+            # @param [Nokogiri::Document] mods_ng_xml MODS to be normalized
+            # @return [Nokogiri::Document] normalized MODS
+            def self.normalize(mods_ng_xml:)
+              new(mods_ng_xml: mods_ng_xml).normalize
+            end
+
+            def initialize(mods_ng_xml:)
+              @ng_xml = mods_ng_xml.dup
+              @ng_xml.encoding = 'UTF-8'
+            end
+
+            def normalize
+              normalize_xlink_href
+              normalize_empty_geographic
+              normalize_marcgac
+              normalize_empty_temporal
+              normalize_subject
+              normalize_subject_children
+              normalize_subject_authority
+              normalize_subject_authority_lcnaf
+              normalize_subject_authority_naf
+              normalize_subject_authority_tgm
+              normalize_coordinates # Must be before normalize_subject_cartographics
+              normalize_subject_cartographics
+              normalize_subject_lang_and_script
+              ng_xml
+            end
+
+            private
+
+            attr_reader :ng_xml
+
+            def normalize_marcgac
+              ng_xml.root.xpath('//mods:geographicCode', mods: ModsNormalizer::MODS_NS).each do |node|
+                node.content = node.content.sub(/-+$/, '')
+              end
+            end
+
+            # rubocop:disable Metrics/CyclomaticComplexity
+            # rubocop:disable Metrics/AbcSize
+            def normalize_subject
+              ng_xml.root.xpath('//mods:subject[not(mods:cartographics)]', mods: ModsNormalizer::MODS_NS).each do |subject_node|
+                children_nodes = subject_node.xpath('mods:*', mods: ModsNormalizer::MODS_NS)
+
+                next if children_nodes.empty?
+
+                if (has_authorityURI?(subject_node) || has_valueURI?(subject_node)) &&
+                   children_nodes.size == 1
+                  # If subject has authority and child doesn't, copy to child.
+                  add_authority(children_nodes, subject_node) if has_authority?(subject_node) && !has_authority?(children_nodes)
+                  # If subject has authorityURI and child doesn't, move to child.
+                  add_authorityURI(children_nodes, subject_node) if has_authorityURI?(subject_node) && !has_authorityURI?(children_nodes)
+                  subject_node.delete('authorityURI')
+                  # If subject has valueURI and child doesn't, move to child.
+                  add_valueURI(children_nodes, subject_node) if has_valueURI?(subject_node) && !has_valueURI?(children_nodes)
+                  subject_node.delete('valueURI')
+                end
+
+                if !has_authority?(subject_node) &&
+                   has_authority?(children_nodes.first) &&
+                   has_same_authority?(children_nodes, children_nodes.first)
+                  add_authority(subject_node, children_nodes.first, naf_to_lcsh: true)
+                end
+
+                next unless has_authority?(subject_node) &&
+                            has_authorityURI?(subject_node) &&
+                            !has_valueURI?(subject_node) &&
+                            has_authority?(children_nodes.first) &&
+                            has_same_authority?(children_nodes, children_nodes.first)
+
+                delete_authorityURI(subject_node)
+              end
+            end
+
+            def normalize_subject_children
+              ng_xml.root.xpath('//mods:subject[not(mods:cartographics)]', mods: ModsNormalizer::MODS_NS).each do |subject_node|
+                children_nodes = subject_node.xpath('mods:*', mods: ModsNormalizer::MODS_NS)
+
+                children_nodes.each do |child_node|
+                  next unless !has_authorityURI?(subject_node) &&
+                              !has_valueURI?(subject_node) &&
+                              has_authority?(child_node) &&
+                              has_same_authority?(child_node, subject_node) &&
+                              child_node['authority'] != 'naf' &&
+                              !(has_authorityURI?(child_node) || has_valueURI?(child_node))
+
+                  delete_authority(child_node)
+                end
+
+                next unless !has_authorityURI?(subject_node) &&
+                            !has_valueURI?(subject_node) &&
+                            has_authority?(subject_node) &&
+                            !has_authority?(children_nodes) &&
+                            (has_authorityURI?(children_nodes) || has_valueURI?(children_nodes))
+
+                add_authority(children_nodes, subject_node)
+              end
+            end
+            # rubocop:enable Metrics/CyclomaticComplexity
+            # rubocop:enable Metrics/AbcSize
+
+            # rubocop:disable Naming/PredicateName
+            def has_authority?(nodes)
+              nodes_to_a(nodes).all? { |node| node[:authority] }
+            end
+
+            def has_same_authority?(nodes, same_node)
+              nodes_to_a(nodes).all? { |node| same_node[:authority] == node[:authority] || (lcsh_or_naf?(same_node) && lcsh_or_naf?(node)) }
+            end
+            # rubocop:enable Naming/PredicateName
+
+            def lcsh_or_naf?(node)
+              %w[lcsh naf].include?(node[:authority])
+            end
+
+            def add_authority(nodes, from_node, naf_to_lcsh: false)
+              authority = if naf_to_lcsh && from_node[:authority] == 'naf'
+                            'lcsh'
+                          else
+                            from_node[:authority]
+                          end
+              nodes_to_a(nodes).each { |node| node[:authority] = authority }
+            end
+
+            def delete_authority(nodes)
+              nodes_to_a(nodes).each { |node| node.delete('authority') }
+            end
+
+            # rubocop:disable Naming/MethodName
+            # rubocop:disable Naming/PredicateName
+            def has_authorityURI?(nodes)
+              nodes_to_a(nodes).all? { |node| node[:authorityURI] }
+            end
+
+            def add_authorityURI(nodes, from_node)
+              nodes_to_a(nodes).each { |node| node[:authorityURI] = from_node[:authorityURI] }
+            end
+
+            def delete_authorityURI(nodes)
+              nodes_to_a(nodes).each { |node| node.delete('authorityURI') }
+            end
+
+            def has_valueURI?(nodes)
+              nodes_to_a(nodes).all? { |node| node[:valueURI] }
+            end
+
+            def add_valueURI(nodes, from_node)
+              nodes_to_a(nodes).each { |node| node[:valueURI] = from_node[:valueURI] }
+            end
+            # rubocop:enable Naming/MethodName
+            # rubocop:enable Naming/PredicateName
+
+            def nodes_to_a(nodes)
+              nodes.is_a?(Nokogiri::XML::NodeSet) ? nodes : [nodes]
+            end
+
+            def normalize_subject_authority
+              ng_xml.root.xpath('//mods:subject[not(@authority) and count(mods:*) = 1 and not(mods:geographicCode)]/mods:*[@authority and @authority != "naf"]',
+                                mods: ModsNormalizer::MODS_NS).each do |node|
+                node.parent['authority'] = node['authority']
+                node.delete('authority') unless node['authorityURI'] || node['valueURI']
+              end
+            end
+
+            def normalize_subject_authority_lcnaf
+              ng_xml.root.xpath("//mods:*[@authority='lcnaf']", mods: ModsNormalizer::MODS_NS).each do |node|
+                node[:authority] = 'naf'
+              end
+            end
+
+            def normalize_subject_authority_tgm
+              ng_xml.root.xpath("//mods:*[@authority='tgm']", mods: ModsNormalizer::MODS_NS).each do |node|
+                node[:authority] = 'lctgm'
+              end
+            end
+
+            def normalize_coordinates
+              ng_xml.root.xpath('//mods:coordinates[text()]', mods: ModsNormalizer::MODS_NS).each do |coordinate_node|
+                coordinate_node.content = coordinate_node.content.delete_prefix('(').delete_suffix(')')
+              end
+            end
+
+            # Collapse multiple subject/cartographics nodes into a single one
+            def normalize_subject_cartographics
+              normalize_subject_cartographics_for(ng_xml.root)
+              ng_xml.root.xpath('mods:relatedItem', mods: ModsNormalizer::MODS_NS).each do |related_item_node|
+                normalize_subject_cartographics_for(related_item_node)
+              end
+            end
+
+            def normalize_subject_cartographics_for(root_node)
+              carto_subject_nodes = root_node.xpath('mods:subject[not(@altRepGroup)][mods:cartographics]', mods: ModsNormalizer::MODS_NS)
+              return if carto_subject_nodes.empty?
+
+              # Create a default carto subject.
+              default_carto_subject_node = Nokogiri::XML::Node.new('subject', Nokogiri::XML(nil))
+              default_carto_node = Nokogiri::XML::Node.new('cartographics', Nokogiri::XML(nil))
+              default_carto_subject_node << default_carto_node
+
+              carto_subject_nodes.each do |carto_subject_node|
+                carto_subject_node.xpath('mods:cartographics', mods: ModsNormalizer::MODS_NS).each do |carto_node|
+                  normalize_cartographic_node(carto_node, carto_subject_node, default_carto_node)
+                end
+                carto_subject_node.remove if carto_subject_node.elements.empty?
+              end
+
+              root_node << default_carto_subject_node if default_carto_node.elements.present?
+            end
+
+            # Normalizes a single cartographic node
+            def normalize_cartographic_node(carto_node, carto_subject_node, default_carto_node)
+              child_nodes = if carto_subject_node['authority'] || carto_subject_node['authorityURI'] || carto_subject_node['valueURI']
+                              # Move scale and coordinates to default carto subject.
+                              carto_node.xpath('mods:scale',
+                                               mods: ModsNormalizer::MODS_NS) + carto_node.xpath('mods:coordinates', mods: ModsNormalizer::MODS_NS)
+                            else
+                              # Merge all into default carto_subject.
+                              carto_node.elements
+                            end
+
+              child_nodes.each do |child_node|
+                child_node.remove
+                next if child_node.children.blank? # skip empty nodes
+
+                default_carto_node << child_node unless child_node_exists?(child_node, default_carto_node)
+              end
+              carto_node.remove if carto_node.elements.empty?
+            end
+
+            def child_node_exists?(child_node, parent_node)
+              parent_node.elements.any? { |check_node| child_node.name == check_node.name && child_node.content == check_node.content }
+            end
+
+            def normalize_subject_authority_naf
+              ng_xml.root.xpath("//mods:subject[@authority='naf']", mods: ModsNormalizer::MODS_NS).each do |subject_node|
+                subject_node[:authority] = 'lcsh'
+              end
+            end
+
+            def normalize_subject_lang_and_script
+              ng_xml.root.xpath('//mods:subject[mods:*]', mods: ModsNormalizer::MODS_NS).each do |subject_node|
+                check_child_node = subject_node.elements.first
+                # If all children have the same lang, then move to subject and delete from children.
+                check_lang = check_child_node['lang']
+                if check_lang && subject_node.elements.all? { |child_node| child_node['lang'] == check_lang }
+                  subject_node['lang'] = check_lang
+                  subject_node.elements.each { |child_node| child_node.delete('lang') }
+                end
+
+                # If all children have the same script, then move to subject and delete from children.
+                check_script = check_child_node['script']
+                if check_script && subject_node.elements.all? { |node| node['script'] == check_script }
+                  subject_node['script'] = check_script
+                  subject_node.elements.each { |child_node| child_node.delete('script') }
+                end
+              end
+            end
+
+            def normalize_empty_temporal
+              ng_xml.root.xpath('//mods:subject/mods:temporal[not(text())]', mods: ModsNormalizer::MODS_NS).each do |temporal_node|
+                subject_node = temporal_node.parent
+                temporal_node.remove
+                subject_node.remove if subject_node.elements.empty?
+              end
+            end
+
+            def normalize_empty_geographic
+              ng_xml.root.xpath('//mods:subject/mods:geographic[not(@valueURI) and not(text())]', mods: ModsNormalizer::MODS_NS).each do |geo_node|
+                subject_node = geo_node.parent
+                geo_node.remove
+                subject_node.remove if subject_node.elements.empty? && subject_node.attributes.empty?
+              end
+            end
+
+            def normalize_xlink_href
+              ng_xml.root.xpath('//mods:subject/mods:*[@xlink:href]', mods: ModsNormalizer::MODS_NS, xlink: ModsNormalizer::XLINK_NS).each do |child_node|
+                subject_node = child_node.parent
+                subject_node['xlink:href'] = child_node['xlink:href']
+                child_node.delete('href')
+              end
+            end
+          end
+        end
+      end
+    end
+  end
+end

data/lib/cocina/models/mapping/normalizers/mods/title_normalizer.rb

@@ -0,0 +1,91 @@
+# frozen_string_literal: true
+
+module Cocina
+  module Models
+    module Mapping
+      module Normalizers
+        module Mods
+          # Normalizes a Fedora MODS document for title elements.
+          class TitleNormalizer
+            # @param [Nokogiri::Document] mods_ng_xml MODS to be normalized
+            # @param [String] label
+            # @return [Nokogiri::Document] normalized MODS
+            def self.normalize(mods_ng_xml:, label:)
+              new(mods_ng_xml: mods_ng_xml, label: label).normalize
+            end
+
+            def self.normalize_missing_title(mods_ng_xml:, label:)
+              new(mods_ng_xml: mods_ng_xml, label: label).normalize_missing_title
+            end
+
+            def initialize(mods_ng_xml:, label:)
+              @ng_xml = mods_ng_xml.dup
+              @ng_xml.encoding = 'UTF-8'
+              @label = label
+            end
+
+            def normalize
+              normalize_hydrus_title
+              clean_empty_titles
+              normalize_title_type
+              normalize_title_trailing
+              normalize_title_as_label
+              ng_xml
+            end
+
+            def normalize_missing_title
+              normalize_title_as_label
+              ng_xml
+            end
+
+            private
+
+            attr_reader :ng_xml, :label
+
+            def normalize_hydrus_title
+              titles = ng_xml.root.xpath('mods:titleInfo/mods:title[string-length() > 0]', mods: ModsNormalizer::MODS_NS)
+              return if titles.present? || label != 'Hydrus'
+
+              add_title('Hydrus')
+            end
+
+            def clean_empty_titles
+              ng_xml.root.xpath('//mods:title[not(text())]', mods: ModsNormalizer::MODS_NS).each(&:remove)
+              ng_xml.root.xpath('//mods:subTitle[not(text())]', mods: ModsNormalizer::MODS_NS).each(&:remove)
+              ng_xml.root.xpath('//mods:titleInfo[not(mods:*) and not(@xlink:href)]',
+                                mods: ModsNormalizer::MODS_NS, xlink: ModsNormalizer::XLINK_NS).each(&:remove)
+            end
+
+            def normalize_title_type
+              ng_xml.root.xpath('//mods:title[@type]', mods: ModsNormalizer::MODS_NS).each do |title_node|
+                title_node.delete('type')
+              end
+            end
+
+            def normalize_title_trailing
+              ng_xml.root.xpath('//mods:titleInfo[not(@type="abbreviated")]/mods:title', mods: ModsNormalizer::MODS_NS).each do |title_node|
+                title_node.content = title_node.content.delete_suffix(',')
+              end
+            end
+
+            def normalize_title_as_label
+              return if ng_xml.root.xpath('mods:titleInfo/mods:title',
+                                          mods: ModsNormalizer::MODS_NS).present? || ng_xml.root.xpath('mods:titleInfo[@xlink:href]', mods: ModsNormalizer::MODS_NS,
+                                                                                                                                      xlink: ModsNormalizer::XLINK_NS).present?
+
+              add_title(label)
+            end
+
+            def add_title(content)
+              new_title_info = Nokogiri::XML::Node.new('titleInfo', Nokogiri::XML(nil))
+              new_title = Nokogiri::XML::Node.new('title', Nokogiri::XML(nil))
+              new_title.content = content
+              new_title_info << new_title
+              ng_xml.root << new_title_info
+            end
+          end
+        end
+      end
+    end
+  end
+end