arclight 0.2.0 → 0.3.0

data/lib/arclight/normalized_date.rb

@@ -47,8 +47,8 @@ module Arclight
       else
         result = nil
       end
-      return if result.blank?
-      result.strip
+
+      result&.strip
     end
   end
 end

data/lib/arclight/normalized_id.rb

@@ -19,6 +19,7 @@ module Arclight
 
     def normalize
       raise Arclight::Exceptions::IDNotFound if id.blank?
+
       id.strip.tr('.', '-')
     end
   end

data/lib/arclight/normalized_title.rb

@@ -24,6 +24,7 @@ module Arclight
     def normalize
       result = [title, date].compact.join(', ')
       raise Arclight::Exceptions::TitleNotFound if result.blank?
+
       result
     end
   end

data/lib/arclight/repository.rb

@@ -26,6 +26,57 @@ module Arclight
       [city, state_zip, country].compact.join(', ')
     end
 
+    # Why are we using self#respond_to? below?
+    #
+    # All the keys in the config hash from `repositories.yml` are
+    # on-the-fly added as attr_accessors up in #initialize. If the
+    # request_types key isn't present, the method won't be created.
+    #
+    # Since the original data is thrown away, this is the best way
+    # to see if that key was present.
+    def request_config_present?
+      return false unless respond_to? :request_types
+      return false if request_types.nil? || request_types.empty?
+
+      request_configs = request_types.map { |_k, v| v }
+      request_configs[0]&.fetch('request_url').present? &&
+        request_configs[0]&.fetch('request_mappings').present?
+    end
+
+    def request_config_present_for_type?(type)
+      return false unless type && request_config_present?
+
+      config = request_types[type]
+      config&.fetch('request_url').present? &&
+        config&.fetch('request_mappings').present?
+    end
+
+    def request_config_for_type(type)
+      return nil unless type && request_config_present_for_type?(type)
+
+      request_types[type]
+    end
+
+    def request_url_for_type(type)
+      return nil unless type && request_config_present_for_type?(type)
+
+      config = request_config_for_type(type)
+      config.fetch('request_url')
+    end
+
+    def request_mappings_for_type(type)
+      return nil unless type && request_config_present_for_type?(type)
+
+      config = request_config_for_type(type)
+      config.fetch('request_mappings')
+    end
+
+    def available_request_types
+      return [] unless request_types.present?
+
+      request_types.keys
+    end
+
     # Load repository information from a YAML file
     #
     # @param [String] `filename`
@@ -42,19 +93,20 @@ module Arclight
     # Mimics ActiveRecord's `all` behavior
     #
     # @return [Array<Repository>]
-    def self.all
-      from_yaml(ENV['REPOSITORY_FILE'] || 'config/repositories.yml').values
+    def self.all(yaml_file = nil)
+      yaml_file = ENV['REPOSITORY_FILE'] || 'config/repositories.yml' if yaml_file.nil?
+      from_yaml(yaml_file).values
     end
 
     # Mimics ActiveRecord dynamic `find_by` behavior for the slug or name
     #
     # @param [String] `slug` or `name`
     # @return [Repository]
-    def self.find_by(slug: nil, name: nil)
+    def self.find_by(slug: nil, name: nil, yaml_file: nil)
       if slug
-        all.find { |repo| repo.slug == slug }
+        all(yaml_file).find { |repo| repo.slug == slug }
       elsif name
-        all.find { |repo| repo.name == name }
+        all(yaml_file).find { |repo| repo.name == name }
       else
         raise ArgumentError, 'Requires either slug or name parameters to find_by'
       end
@@ -68,6 +120,7 @@ module Arclight
     def self.find_by!(*args)
       repository = find_by(*args)
       raise ActiveRecord::RecordNotFound if repository.blank?
+
       repository
     end
   end

data/lib/arclight/traject/ead2_config.rb

@@ -5,6 +5,7 @@ require 'traject'
 require 'traject/nokogiri_reader'
 require 'traject_plus'
 require 'traject_plus/macros'
+require 'arclight/level_label'
 require 'arclight/normalized_date'
 require 'arclight/normalized_title'
 require 'active_model/conversion' ## Needed for Arclight::Repository
@@ -13,13 +14,14 @@ require 'arclight/digital_object'
 require 'arclight/year_range'
 require 'arclight/repository'
 require 'arclight/missing_id_strategy'
-
-NAME_ELEMENTS = %w[corpname famname name persname].freeze
+require 'arclight/traject/nokogiri_namespaceless_reader'
 
 # rubocop:disable Style/MixinUsage
 extend TrajectPlus::Macros
 # rubocop:enable Style/MixinUsage
 
+NAME_ELEMENTS = %w[corpname famname name persname].freeze
+
 SEARCHABLE_NOTES_FIELDS = %w[
   accessrestrict
   accruals
@@ -50,8 +52,7 @@ DID_SEARCHABLE_NOTES_FIELDS = %w[
 ].freeze
 
 settings do
-  provide 'nokogiri.namespaces',
-          'xmlns' => 'urn:isbn:1-931666-22-9'
+  provide 'reader_class_name', 'Arclight::Traject::NokogiriNamespacelessReader'
   provide 'solr_writer.commit_on_close', 'true'
   provide 'repository', ENV['REPOSITORY_ID']
   provide 'logger', Logger.new($stderr)
@@ -65,28 +66,38 @@ each_record do |_record, context|
   ).name
 end
 
-# Top level
-to_field 'id', extract_xpath('//xmlns:eadid'), strip, gsub('.', '-')
-to_field 'title_filing_si', extract_xpath('//xmlns:titleproper[@type="filing"]')
-to_field 'title_ssm', extract_xpath('//xmlns:archdesc/xmlns:did/xmlns:unittitle')
-to_field 'title_teim', extract_xpath('//xmlns:archdesc/xmlns:did/xmlns:unittitle')
-to_field 'ead_ssi', extract_xpath('//xmlns:eadid')
+# ==================
+# Top level document
+# ==================
 
-to_field 'unitdate_ssm', extract_xpath('//xmlns:archdesc/xmlns:did/xmlns:unitdate')
-to_field 'unitdate_bulk_ssim', extract_xpath('//xmlns:archdesc/xmlns:did/xmlns:unitdate[@type="bulk"]')
-to_field 'unitdate_inclusive_ssm', extract_xpath('//xmlns:archdesc/xmlns:did/xmlns:unitdate[@type="inclusive"]')
-to_field 'unitdate_other_ssim', extract_xpath('//xmlns:archdesc/xmlns:did/xmlns:unitdate[not(@type)]')
+to_field 'id', extract_xpath('/ead/eadheader/eadid'), strip, gsub('.', '-')
+to_field 'title_filing_si', extract_xpath('/ead/eadheader/filedesc/titlestmt/titleproper[@type="filing"]')
+to_field 'title_ssm', extract_xpath('/ead/archdesc/did/unittitle')
+to_field 'title_teim', extract_xpath('/ead/archdesc/did/unittitle')
+to_field 'ead_ssi', extract_xpath('/ead/eadheader/eadid')
 
-to_field 'level_ssm' do |record, accumulator|
-  accumulator << record.at_xpath('//xmlns:archdesc').attribute('level').value
+to_field 'unitdate_ssm', extract_xpath('/ead/archdesc/did/unitdate')
+to_field 'unitdate_bulk_ssim', extract_xpath('/ead/archdesc/did/unitdate[@type="bulk"]')
+to_field 'unitdate_inclusive_ssm', extract_xpath('/ead/archdesc/did/unitdate[@type="inclusive"]')
+to_field 'unitdate_other_ssim', extract_xpath('/ead/archdesc/did/unitdate[not(@type)]')
+
+# All top-level docs treated as 'collection' for routing / display purposes
+to_field 'level_ssm' do |_record, accumulator|
+  accumulator << 'collection'
 end
 
+# Keep the original top-level archdesc/@level for Level facet in addition to 'Collection'
 to_field 'level_sim' do |record, accumulator|
-  accumulator << record.at_xpath('//xmlns:archdesc').attribute('level').value&.capitalize
+  level = record.at_xpath('/ead/archdesc').attribute('level')&.value
+  other_level = record.at_xpath('/ead/archdesc').attribute('otherlevel')&.value
+
+  accumulator << Arclight::LevelLabel.new(level, other_level).to_s
+  accumulator << 'Collection' unless level == 'collection'
 end
 
-to_field 'unitid_ssm', extract_xpath('//xmlns:archdesc/xmlns:did/xmlns:unitid')
-to_field 'unitid_teim', extract_xpath('//xmlns:archdesc/xmlns:did/xmlns:unitid')
+to_field 'unitid_ssm', extract_xpath('/ead/archdesc/did/unitid')
+to_field 'unitid_teim', extract_xpath('/ead/archdesc/did/unitid')
+to_field 'collection_unitid_ssm', extract_xpath('/ead/archdesc/did/unitid')
 
 to_field 'normalized_title_ssm' do |_record, accumulator, context|
   dates = Arclight::NormalizedDate.new(
@@ -112,6 +123,12 @@ end
 to_field 'collection_sim' do |_record, accumulator, context|
   accumulator.concat context.output_hash.fetch('normalized_title_ssm', [])
 end
+to_field 'collection_ssi' do |_record, accumulator, context|
+  accumulator.concat context.output_hash.fetch('normalized_title_ssm', [])
+end
+to_field 'collection_title_tesim' do |_record, accumulator, context|
+  accumulator.concat context.output_hash.fetch('normalized_title_ssm', [])
+end
 
 to_field 'repository_ssm' do |_record, accumulator, context|
   accumulator << context.clipboard[:repository]
@@ -121,26 +138,25 @@ to_field 'repository_sim' do |_record, accumulator, context|
   accumulator << context.clipboard[:repository]
 end
 
-to_field 'geogname_ssm', extract_xpath('//xmlns:archdesc/xmlns:controlaccess/xmlns:geogname')
-
-to_field 'geogname_sim', extract_xpath('//xmlns:archdesc/xmlns:controlaccess/xmlns:geogname')
+to_field 'geogname_ssm', extract_xpath('/ead/archdesc/controlaccess/geogname')
+to_field 'geogname_sim', extract_xpath('/ead/archdesc/controlaccess/geogname')
 
-to_field 'creator_ssm', extract_xpath("//xmlns:archdesc/xmlns:did/xmlns:origination[@label='creator']")
-to_field 'creator_sim', extract_xpath("//xmlns:archdesc/xmlns:did/xmlns:origination[@label='creator']")
-to_field 'creator_ssim', extract_xpath("//xmlns:archdesc/xmlns:did/xmlns:origination[@label='creator']")
+to_field 'creator_ssm', extract_xpath('/ead/archdesc/did/origination')
+to_field 'creator_sim', extract_xpath('/ead/archdesc/did/origination')
+to_field 'creator_ssim', extract_xpath('/ead/archdesc/did/origination')
 to_field 'creator_sort' do |record, accumulator|
-  accumulator << record.xpath("//xmlns:archdesc/xmlns:did/xmlns:origination[@label='creator']").map { |c| c.text.strip }.join(', ')
+  accumulator << record.xpath('/ead/archdesc/did/origination').map { |c| c.text.strip }.join(', ')
 end
 
-to_field 'creator_persname_ssm', extract_xpath("//xmlns:archdesc/xmlns:did/xmlns:origination[@label='creator']/xmlns:persname")
-to_field 'creator_persname_ssim', extract_xpath("//xmlns:archdesc/xmlns:did/xmlns:origination[@label='creator']/xmlns:persname")
-to_field 'creator_corpname_ssm', extract_xpath("//xmlns:archdesc/xmlns:did/xmlns:origination[@label='creator']/xmlns:corpname")
-to_field 'creator_corpname_sim', extract_xpath("//xmlns:archdesc/xmlns:did/xmlns:origination[@label='creator']/xmlns:corpname")
-to_field 'creator_corpname_ssim', extract_xpath("//xmlns:archdesc/xmlns:did/xmlns:origination[@label='creator']/xmlns:corpname")
-to_field 'creator_famname_ssm', extract_xpath("//xmlns:archdesc/xmlns:did/xmlns:origination[@label='creator']/xmlns:famname")
-to_field 'creator_famname_ssim', extract_xpath("//xmlns:archdesc/xmlns:did/xmlns:origination[@label='creator']/xmlns:famname")
+to_field 'creator_persname_ssm', extract_xpath('/ead/archdesc/did/origination/persname')
+to_field 'creator_persname_ssim', extract_xpath('/ead/archdesc/did/origination/persname')
+to_field 'creator_corpname_ssm', extract_xpath('/ead/archdesc/did/origination/corpname')
+to_field 'creator_corpname_sim', extract_xpath('/ead/archdesc/did/origination/corpname')
+to_field 'creator_corpname_ssim', extract_xpath('/ead/archdesc/did/origination/corpname')
+to_field 'creator_famname_ssm', extract_xpath('/ead/archdesc/did/origination/famname')
+to_field 'creator_famname_ssim', extract_xpath('/ead/archdesc/did/origination/famname')
 
-to_field 'persname_sim', extract_xpath('//xmlns:persname')
+to_field 'persname_sim', extract_xpath('//persname')
 
 to_field 'creators_ssim' do |_record, accumulator, context|
   accumulator.concat context.output_hash['creator_persname_ssm'] if context.output_hash['creator_persname_ssm']
@@ -148,27 +164,19 @@ to_field 'creators_ssim' do |_record, accumulator, context|
   accumulator.concat context.output_hash['creator_famname_ssm'] if context.output_hash['creator_famname_ssm']
 end
 
-to_field 'places_sim', extract_xpath('//xmlns:archdesc/xmlns:controlaccess/xmlns:geogname')
-to_field 'places_ssim', extract_xpath('//xmlns:archdesc/xmlns:controlaccess/xmlns:geogname')
-to_field 'places_ssm', extract_xpath('//xmlns:archdesc/xmlns:controlaccess/xmlns:geogname')
+to_field 'places_sim', extract_xpath('/ead/archdesc/controlaccess/geogname')
+to_field 'places_ssim', extract_xpath('/ead/archdesc/controlaccess/geogname')
+to_field 'places_ssm', extract_xpath('/ead/archdesc/controlaccess/geogname')
 
-to_field 'access_terms_ssm', extract_xpath('//xmlns:archdesc/xmlns:userestrict/xmlns:p')
+to_field 'access_terms_ssm', extract_xpath('/ead/archdesc/userestrict/*[local-name()!="head"]')
 
-# Indexes the acquisition group information into the notes field
-# Please see https://www.loc.gov/ead/tglib/elements/acqinfo.html
-to_field 'acqinfo_ssim', extract_xpath('/xmlns:ead/xmlns:archdesc/xmlns:acqinfo/*[local-name()!="head"]')
-to_field 'acqinfo_ssim', extract_xpath('/xmlns:ead/xmlns:archdesc/xmlns:descgrp/xmlns:acqinfo/*[local-name()!="head"]')
-to_field 'acqinfo_ssim', extract_xpath('./xmlns:acqinfo/*[local-name()!="head"]')
-to_field 'acqinfo_ssim', extract_xpath('./xmlns:descgrp/xmlns:acqinfo/*[local-name()!="head"]')
-to_field 'acqinfo_ssm' do |_record, accumulator, context|
-  accumulator.concat(context.output_hash.fetch('acqinfo_ssim', []))
-end
+to_field 'acqinfo_ssim', extract_xpath('/ead/archdesc/acqinfo/*[local-name()!="head"]')
+to_field 'acqinfo_ssim', extract_xpath('/ead/archdesc/descgrp/acqinfo/*[local-name()!="head"]')
 
-# Indexes only specified controlled terms for archival description into the access_subject field
-to_field 'access_subjects_ssim', extract_xpath('//xmlns:archdesc/xmlns:controlaccess', to_text: false) do |_record, accumulator|
+to_field 'access_subjects_ssim', extract_xpath('/ead/archdesc/controlaccess', to_text: false) do |_record, accumulator|
   accumulator.map! do |element|
     %w[subject function occupation genreform].map do |selector|
-      element.xpath(".//xmlns:#{selector}").map(&:text)
+      element.xpath(".//#{selector}").map(&:text)
     end
   end.flatten!
 end
@@ -177,16 +185,25 @@ to_field 'access_subjects_ssm' do |_record, accumulator, context|
   accumulator.concat Array.wrap(context.output_hash['access_subjects_ssim'])
 end
 
-to_field 'has_online_content_ssim', extract_xpath('.//xmlns:dao') do |_record, accumulator|
+to_field 'has_online_content_ssim', extract_xpath('.//dao') do |_record, accumulator|
   accumulator.replace([accumulator.any?])
 end
 
-to_field 'extent_ssm', extract_xpath('//xmlns:did/xmlns:physdesc/xmlns:extent')
-to_field 'extent_teim', extract_xpath('//xmlns:did/xmlns:physdesc/xmlns:extent')
-to_field 'genreform_sim', extract_xpath('//xmlns:archdesc/xmlns:controlaccess/xmlns:genreform')
-to_field 'genreform_ssm', extract_xpath('//xmlns:archdesc/xmlns:controlaccess/xmlns:genreform')
+to_field 'digital_objects_ssm', extract_xpath('/ead/archdesc/did/dao|/ead/archdesc/dao', to_text: false) do |_record, accumulator|
+  accumulator.map! do |dao|
+    label = dao.attributes['title']&.value ||
+            dao.xpath('daodesc/p')&.text
+    href = (dao.attributes['href'] || dao.attributes['xlink:href'])&.value
+    Arclight::DigitalObject.new(label: label, href: href).to_json
+  end
+end
 
-to_field 'date_range_sim', extract_xpath('.//xmlns:did/xmlns:unitdate/@normal', to_text: false) do |_record, accumulator|
+to_field 'extent_ssm', extract_xpath('/ead/archdesc/did/physdesc/extent')
+to_field 'extent_teim', extract_xpath('/ead/archdesc/did/physdesc/extent')
+to_field 'genreform_sim', extract_xpath('/ead/archdesc/controlaccess/genreform')
+to_field 'genreform_ssm', extract_xpath('/ead/archdesc/controlaccess/genreform')
+
+to_field 'date_range_sim', extract_xpath('/ead/archdesc/did/unitdate/@normal', to_text: false) do |_record, accumulator|
   range = Arclight::YearRange.new
   next range.years if accumulator.blank?
 
@@ -196,26 +213,33 @@ to_field 'date_range_sim', extract_xpath('.//xmlns:did/xmlns:unitdate/@normal',
 end
 
 SEARCHABLE_NOTES_FIELDS.map do |selector|
-  to_field "#{selector}_ssm", extract_xpath("//xmlns:archdesc/xmlns:#{selector}/*[local-name()!='head']")
-  to_field "#{selector}_heading_ssm", extract_xpath("//xmlns:archdesc/xmlns:#{selector}/xmlns:head") unless selector == 'prefercite'
-  to_field "#{selector}_teim", extract_xpath("//xmlns:archdesc/xmlns:#{selector}/*[local-name()!='head']")
+  to_field "#{selector}_ssm", extract_xpath("/ead/archdesc/#{selector}/*[local-name()!='head']", to_text: false)
+  to_field "#{selector}_heading_ssm", extract_xpath("/ead/archdesc/#{selector}/head") unless selector == 'prefercite'
+  to_field "#{selector}_teim", extract_xpath("/ead/archdesc/#{selector}/*[local-name()!='head']")
 end
 
 DID_SEARCHABLE_NOTES_FIELDS.map do |selector|
-  to_field "#{selector}_ssm", extract_xpath("//xmlns:did/xmlns:#{selector}")
+  to_field "#{selector}_ssm", extract_xpath("/ead/archdesc/did/#{selector}", to_text: false)
 end
+
 NAME_ELEMENTS.map do |selector|
-  to_field 'names_coll_ssim', extract_xpath("/xmlns:ead/xmlns:archdesc/xmlns:controlaccess/xmlns:#{selector}")
-  to_field 'names_ssim', extract_xpath("//xmlns:#{selector}")
-  to_field "#{selector}_ssm", extract_xpath("//xmlns:#{selector}")
+  to_field 'names_coll_ssim', extract_xpath("/ead/archdesc/controlaccess/#{selector}")
+  to_field 'names_ssim', extract_xpath("//#{selector}")
+  to_field "#{selector}_ssm", extract_xpath("//#{selector}")
 end
-to_field 'corpname_sim', extract_xpath('//xmlns:corpname')
 
-to_field 'language_sim', extract_xpath('//xmlns:did/xmlns:langmaterial')
-to_field 'language_ssm', extract_xpath('//xmlns:did/xmlns:langmaterial')
+to_field 'corpname_sim', extract_xpath('//corpname')
+
+to_field 'language_sim', extract_xpath('/ead/archdesc/did/langmaterial')
+to_field 'language_ssm', extract_xpath('/ead/archdesc/did/langmaterial')
+
+to_field 'descrules_ssm', extract_xpath('/ead/eadheader/profiledesc/descrules')
 
+# =============================
 # Each component child document
 # <c> <c01> <c12>
+# =============================
+
 compose 'components', ->(record, accumulator, _context) { accumulator.concat record.xpath('//*[is_component(.)]', NokogiriXpathExtensions.new) } do
   to_field 'ref_ssi' do |record, accumulator, context|
     accumulator << if record.attribute('id').blank?
@@ -224,10 +248,12 @@ compose 'components', ->(record, accumulator, _context) { accumulator.concat rec
                      parent_id = context.clipboard[:parent].output_hash['id'].first
                      logger.warn('MISSING ID WARNING') do
                        [
-                         "A component in #{parent_id} did not have and ID so one was minted using the #{strategy} strategy.",
+                         "A component in #{parent_id} did not have an ID so one was minted using the #{strategy} strategy.",
                          "The ID of this document will be #{parent_id}#{hexdigest}."
                        ].join(' ')
                      end
+                     record['id'] = hexdigest
+                     hexdigest
                    else
                      record.attribute('id')&.value&.strip&.gsub('.', '-')
                    end
@@ -247,13 +273,13 @@ compose 'components', ->(record, accumulator, _context) { accumulator.concat rec
     accumulator << context.clipboard[:parent].output_hash['ead_ssi'].first
   end
 
-  to_field 'title_filing_si', extract_xpath('./xmlns:did/xmlns:unittitle'), first_only
-  to_field 'title_ssm', extract_xpath('./xmlns:did/xmlns:unittitle')
-  to_field 'title_teim', extract_xpath('./xmlns:did/xmlns:unittitle')
+  to_field 'title_filing_si', extract_xpath('./did/unittitle'), first_only
+  to_field 'title_ssm', extract_xpath('./did/unittitle')
+  to_field 'title_teim', extract_xpath('./did/unittitle')
 
-  to_field 'unitdate_bulk_ssim', extract_xpath('./xmlns:did/xmlns:unitdate[@type="bulk"]')
-  to_field 'unitdate_inclusive_ssm', extract_xpath('./xmlns:did/xmlns:unitdate[@type="inclusive"]')
-  to_field 'unitdate_other_ssim', extract_xpath('./xmlns:did/xmlns:unitdate[not(@type)]')
+  to_field 'unitdate_bulk_ssim', extract_xpath('./did/unitdate[@type="bulk"]')
+  to_field 'unitdate_inclusive_ssm', extract_xpath('./did/unitdate[@type="inclusive"]')
+  to_field 'unitdate_other_ssim', extract_xpath('./did/unitdate[not(@type)]')
 
   to_field 'normalized_title_ssm' do |_record, accumulator, context|
     dates = Arclight::NormalizedDate.new(
@@ -274,33 +300,50 @@ compose 'components', ->(record, accumulator, _context) { accumulator.concat rec
   end
 
   to_field 'component_level_isim' do |record, accumulator|
-    accumulator << 1 + record.ancestors.count { |node| node.name == 'c' }
+    accumulator << 1 + NokogiriXpathExtensions.new.is_component(record.ancestors).count
   end
 
-  to_field 'parent_ssm' do |record, accumulator, context|
+  to_field 'parent_ssim' do |record, accumulator, context|
     accumulator << context.clipboard[:parent].output_hash['id'].first
-    accumulator.concat NokogiriXpathExtensions.new.is_component(record.ancestors).reverse.map { |n| n.attribute('id').value }
+    accumulator.concat NokogiriXpathExtensions.new.is_component(record.ancestors).reverse.map { |n| n.attribute('id')&.value }
   end
 
   to_field 'parent_ssi' do |_record, accumulator, context|
-    accumulator << context.output_hash['parent_ssm'].last
+    accumulator << context.output_hash['parent_ssim'].last
   end
 
-  to_field 'parent_unittitles_ssm' do |_record, accumulator, context|
-    ## Top level document
+  to_field 'parent_unittitles_ssm' do |_rec, accumulator, context|
+    # top level document
     accumulator.concat context.clipboard[:parent].output_hash['normalized_title_ssm']
-    ## Other components
-    context.output_hash['parent_ssm']&.drop(1)&.each do |id|
-      accumulator.concat Array
-        .wrap(context.clipboard[:parent].output_hash['components'])
-        .find { |c| c['ref_ssi'] == [id] }&.[]('normalized_title_ssm')
+    parent_ssim = context.output_hash['parent_ssim']
+    components = context.clipboard[:parent].output_hash['components']
+
+    # other components
+    if parent_ssim && components
+      ancestors = parent_ssim.drop(1).map { |x| [x] }
+      accumulator.concat components.select { |c| ancestors.include? c['ref_ssi'] }.flat_map { |c| c['normalized_title_ssm'] }
     end
   end
+
   to_field 'parent_unittitles_teim' do |_record, accumulator, context|
     accumulator.concat context.output_hash['parent_unittitles_ssm']
   end
 
-  to_field 'unitid_ssm', extract_xpath('./xmlns:did/xmlns:unitid')
+  to_field 'parent_levels_ssm' do |_record, accumulator, context|
+    ## Top level document
+    accumulator.concat context.clipboard[:parent].output_hash['level_ssm']
+    ## Other components
+    context.output_hash['parent_ssim']&.drop(1)&.each do |id|
+      accumulator.concat Array
+        .wrap(context.clipboard[:parent].output_hash['components'])
+        .select { |c| c['ref_ssi'] == [id] }.map { |c| c['level_ssm'] }.flatten
+    end
+  end
+
+  to_field 'unitid_ssm', extract_xpath('./did/unitid')
+  to_field 'collection_unitid_ssm' do |_record, accumulator, context|
+    accumulator.concat Array.wrap(context.clipboard[:parent].output_hash['unitid_ssm'])
+  end
   to_field 'repository_ssm' do |_record, accumulator, context|
     accumulator << context.clipboard[:parent].clipboard[:repository]
   end
@@ -313,22 +356,27 @@ compose 'components', ->(record, accumulator, _context) { accumulator.concat rec
   to_field 'collection_sim' do |_record, accumulator, context|
     accumulator.concat context.clipboard[:parent].output_hash['normalized_title_ssm']
   end
+  to_field 'collection_ssi' do |_record, accumulator, context|
+    accumulator.concat context.clipboard[:parent].output_hash['normalized_title_ssm']
+  end
+
+  to_field 'extent_ssm', extract_xpath('./did/physdesc/extent')
+  to_field 'extent_teim', extract_xpath('./did/physdesc/extent')
 
-  to_field 'extent_ssm', extract_xpath('./xmlns:did/xmlns:physdesc/xmlns:extent')
-  to_field 'creator_ssm', extract_xpath("./xmlns:did/xmlns:origination[@label='creator']")
-  to_field 'creator_ssim', extract_xpath("./xmlns:did/xmlns:origination[@label='creator']")
-  to_field 'creators_ssim', extract_xpath("./xmlns:did/xmlns:origination[@label='creator']")
+  to_field 'creator_ssm', extract_xpath('./did/origination')
+  to_field 'creator_ssim', extract_xpath('./did/origination')
+  to_field 'creators_ssim', extract_xpath('./did/origination')
   to_field 'creator_sort' do |record, accumulator|
-    accumulator << record.xpath("./xmlns:did/xmlns:origination[@label='creator']").map(&:text).join(', ')
+    accumulator << record.xpath('./did/origination').map(&:text).join(', ')
   end
   to_field 'collection_creator_ssm' do |_record, accumulator, context|
     accumulator.concat Array.wrap(context.clipboard[:parent].output_hash['creator_ssm'])
   end
-  to_field 'has_online_content_ssim', extract_xpath('.//xmlns:dao') do |_record, accumulator|
+  to_field 'has_online_content_ssim', extract_xpath('.//dao') do |_record, accumulator|
     accumulator.replace([accumulator.any?])
   end
-  to_field 'child_component_count_isim', extract_xpath('xmlns:c') do |_record, accumulator|
-    accumulator.replace([accumulator.length])
+  to_field 'child_component_count_isim' do |record, accumulator|
+    accumulator << NokogiriXpathExtensions.new.is_component(record.children).count
   end
 
   to_field 'ref_ssm' do |record, accumulator|
@@ -338,13 +386,7 @@ compose 'components', ->(record, accumulator, _context) { accumulator.concat rec
   to_field 'level_ssm' do |record, accumulator|
     level = record.attribute('level')&.value
     other_level = record.attribute('otherlevel')&.value
-
-    accumulator << if level == 'otherlevel'
-                     alternative_level = other_level if other_level
-                     alternative_level.present? ? alternative_level : 'Other'
-                   elsif level.present?
-                     level&.capitalize
-                   end
+    accumulator << Arclight::LevelLabel.new(level, other_level).to_s
   end
 
   to_field 'level_sim' do |_record, accumulator, context|
@@ -353,52 +395,34 @@ compose 'components', ->(record, accumulator, _context) { accumulator.concat rec
     accumulator.concat context.output_hash['level_ssm']&.map(&:capitalize)
   end
 
-  to_field 'parent_access_restrict_ssm', extract_xpath('./xmlns:accessrestrict/xmlns:p')
-
-  to_field 'parent_access_restrict_ssm' do |_record, accumulator, context|
-    next unless context.output_hash['accessrestrict_ssm'].nil?
-
-    context.output_hash['parent_ssm']&.each do |id|
-      accumulator.concat Array
-        .wrap(context.clipboard[:parent]&.output_hash&.[]('components'))
-        .select { |c| c['ref_ssi'] == [id] }.map { |c| c['accessrestrict_ssm'] }
-    end
-  end
-
-  to_field 'parent_access_restrict_ssm' do |_record, accumulator, context|
-    next unless context.output_hash['parent_access_restrict_ssm'].nil?
-
-    accumulator.concat Array.wrap(context.clipboard[:parent]&.output_hash&.[]('accessrestrict_ssm'))
+  to_field 'sort_ii' do |_record, accumulator, context|
+    accumulator.replace([context.position])
   end
 
-  to_field 'parent_access_terms_ssm', extract_xpath('xmlns:userestrict/xmlns:p')
-
-  to_field 'parent_access_terms_ssm' do |_record, accumulator, context|
-    next unless context.output_hash['userestrict_ssm'].nil?
-
-    context.output_hash['parent_ssm']&.each do |id|
-      accumulator.concat Array
-        .wrap(context.clipboard[:parent]&.output_hash&.[]('components'))
-        .select { |c| c['ref_ssi'] == [id] }.map { |c| c['userestrict_ssm'] }
-    end
+  # Get the <accessrestrict> from the closest ancestor that has one (includes top-level)
+  to_field 'parent_access_restrict_ssm' do |record, accumulator|
+    accumulator.concat Array
+      .wrap(record.xpath('(./ancestor::*/accessrestrict)[last()]/*[local-name()!="head"]')
+      .map(&:text))
   end
 
-  to_field 'parent_access_terms_ssm' do |_record, accumulator, context|
-    next unless context.output_hash['parent_access_terms_ssm'].nil?
-
-    accumulator << context.clipboard[:parent]&.output_hash&.[]('access_terms_ssm')&.first
+  # Get the <userestrict> from self OR the closest ancestor that has one (includes top-level)
+  to_field 'parent_access_terms_ssm' do |record, accumulator|
+    accumulator.concat Array
+      .wrap(record.xpath('(./ancestor-or-self::*/userestrict)[last()]/*[local-name()!="head"]')
+      .map(&:text))
   end
 
-  to_field 'digital_objects_ssm', extract_xpath('./xmlns:dao') do |record, accumulator|
-    accumulator.concat(record.xpath('.//xmlns:dao', xmlns: 'urn:isbn:1-931666-22-9').map do |dao|
+  to_field 'digital_objects_ssm', extract_xpath('./dao|./did/dao', to_text: false) do |_record, accumulator|
+    accumulator.map! do |dao|
       label = dao.attributes['title']&.value ||
-        dao.xpath('xmlns:daodesc/xmlns:p', xmlns: 'urn:isbn:1-931666-22-9')&.text
+              dao.xpath('daodesc/p')&.text
       href = (dao.attributes['href'] || dao.attributes['xlink:href'])&.value
       Arclight::DigitalObject.new(label: label, href: href).to_json
-    end.to_a)
+    end
   end
 
-  to_field 'date_range_sim', extract_xpath('.//xmlns:did/xmlns:unitdate/@normal', to_text: false) do |_record, accumulator|
+  to_field 'date_range_sim', extract_xpath('./did/unitdate/@normal', to_text: false) do |_record, accumulator|
     range = Arclight::YearRange.new
     next range.years if accumulator.blank?
 
@@ -408,19 +432,18 @@ compose 'components', ->(record, accumulator, _context) { accumulator.concat rec
   end
 
   NAME_ELEMENTS.map do |selector|
-    to_field 'names_ssim', extract_xpath("./xmlns:controlaccess/xmlns:#{selector}")
-    to_field "#{selector}_ssm", extract_xpath(".//xmlns:#{selector}")
+    to_field 'names_ssim', extract_xpath("./controlaccess/#{selector}")
+    to_field "#{selector}_ssm", extract_xpath(".//#{selector}")
   end
 
-  to_field 'geogname_sim', extract_xpath('./xmlns:controlaccess/xmlns:geogname')
-  to_field 'geogname_ssm', extract_xpath('./xmlns:controlaccess/xmlns:geogname')
-  to_field 'places_ssim', extract_xpath('xmlns:controlaccess/xmlns:geogname')
+  to_field 'geogname_sim', extract_xpath('./controlaccess/geogname')
+  to_field 'geogname_ssm', extract_xpath('./controlaccess/geogname')
+  to_field 'places_ssim', extract_xpath('./controlaccess/geogname')
 
-  # Indexes only specified controlled terms for archival description into the access_subject field
-  to_field 'access_subjects_ssim', extract_xpath('./xmlns:controlaccess', to_text: false) do |_record, accumulator|
+  to_field 'access_subjects_ssim', extract_xpath('./controlaccess', to_text: false) do |_record, accumulator|
     accumulator.map! do |element|
       %w[subject function occupation genreform].map do |selector|
-        element.xpath(".//xmlns:#{selector}").map(&:text)
+        element.xpath(".//#{selector}").map(&:text)
       end
     end.flatten!
   end
@@ -429,31 +452,27 @@ compose 'components', ->(record, accumulator, _context) { accumulator.concat rec
     accumulator.concat(context.output_hash.fetch('access_subjects_ssim', []))
   end
 
-  # Indexes the acquisition group information into the notes field
-  # Please see https://www.loc.gov/ead/tglib/elements/acqinfo.html
-  to_field 'acqinfo_ssim', extract_xpath('/xmlns:ead/xmlns:archdesc/xmlns:acqinfo/*[local-name()!="head"]')
-  to_field 'acqinfo_ssim', extract_xpath('/xmlns:ead/xmlns:archdesc/xmlns:descgrp/xmlns:acqinfo/*[local-name()!="head"]')
-  to_field 'acqinfo_ssim', extract_xpath('./xmlns:acqinfo/*[local-name()!="head"]')
-  to_field 'acqinfo_ssim', extract_xpath('./xmlns:descgrp/xmlns:acqinfo/*[local-name()!="head"]')
-  to_field 'acqinfo_ssm' do |_record, accumulator, context|
-    accumulator.concat(context.output_hash.fetch('acqinfo_ssim', []))
-  end
+  to_field 'acqinfo_ssim', extract_xpath('/ead/archdesc/acqinfo/*[local-name()!="head"]')
+  to_field 'acqinfo_ssim', extract_xpath('/ead/archdesc/descgrp/acqinfo/*[local-name()!="head"]')
+  to_field 'acqinfo_ssim', extract_xpath('./acqinfo/*[local-name()!="head"]')
+  to_field 'acqinfo_ssim', extract_xpath('./descgrp/acqinfo/*[local-name()!="head"]')
 
-  to_field 'language_ssm', extract_xpath('./xmlns:did/xmlns:langmaterial')
+  to_field 'language_ssm', extract_xpath('./did/langmaterial')
   to_field 'containers_ssim' do |record, accumulator|
-    record.xpath('./xmlns:did/xmlns:container').each do |node|
+    record.xpath('./did/container').each do |node|
       accumulator << [node.attribute('type'), node.text].join(' ').strip
     end
   end
+
   SEARCHABLE_NOTES_FIELDS.map do |selector|
-    to_field "#{selector}_ssm", extract_xpath(".//xmlns:#{selector}/*[local-name()!='head']")
-    to_field "#{selector}_heading_ssm", extract_xpath(".//xmlns:archdesc/xmlns:#{selector}/xmlns:head")
-    to_field "#{selector}_teim", extract_xpath(".//xmlns:#{selector}/*[local-name()!='head']")
+    to_field "#{selector}_ssm", extract_xpath("./#{selector}/*[local-name()!='head']", to_text: false)
+    to_field "#{selector}_heading_ssm", extract_xpath("./#{selector}/head")
+    to_field "#{selector}_teim", extract_xpath("./#{selector}/*[local-name()!='head']")
   end
   DID_SEARCHABLE_NOTES_FIELDS.map do |selector|
-    to_field "#{selector}_ssm", extract_xpath(".//xmlns:did/xmlns:#{selector}")
+    to_field "#{selector}_ssm", extract_xpath("./did/#{selector}", to_text: false)
   end
-  to_field 'did_note_ssm', extract_xpath('.//xmlns:did/xmlns:note')
+  to_field 'did_note_ssm', extract_xpath('./did/note')
 end
 
 each_record do |_record, context|