cocina-models 0.74.1 → 0.77.0

data/lib/cocina/models/mapping/from_mods/identifier_builder.rb

@@ -0,0 +1,71 @@
+# frozen_string_literal: true
+
+module Cocina
+  module Models
+    module Mapping
+      module FromMods
+        # Builds cocina identifier
+        class IdentifierBuilder
+          # @param [Nokogiri::XML::Element] identifier_element identifier element
+          # @return [Hash] a hash that can be mapped to a cocina model
+          def self.build_from_identifier(identifier_element:)
+            new(identifier_element: identifier_element, type: identifier_element[:type]).build
+          end
+
+          # @param [Nokogiri::XML::Element] identifier_element recordIdentifier element
+          # @return [Hash] a hash that can be mapped to a cocina model
+          def self.build_from_record_identifier(identifier_element:)
+            new(identifier_element: identifier_element, type: identifier_element[:source]).build
+          end
+
+          # @param [Nokogiri::XML::Element] identifier_element nameIdentifier element
+          # @return [Hash] a hash that can be mapped to a cocina model
+          def self.build_from_name_identifier(identifier_element:)
+            new(identifier_element: identifier_element, type: identifier_element[:type]).build
+          end
+
+          def initialize(identifier_element:, type:)
+            @identifier_element = identifier_element
+            @with_note = with_note
+            @cocina_type, @mods_type, = types_for(type)
+          end
+
+          def build
+            return if identifier_element.text.blank? && identifier_element.attributes.size.zero?
+
+            {
+              displayLabel: identifier_element['displayLabel']
+            }.tap do |attrs|
+              if cocina_type == 'uri'
+                attrs[:uri] = identifier_element.text
+              else
+                attrs[:type] = cocina_type
+                attrs[:value] = identifier_element.text
+                attrs[:source] = build_source
+              end
+              attrs[:status] = 'invalid' if identifier_element['invalid'] == 'yes'
+            end.compact
+          end
+
+          private
+
+          attr_reader :identifier_element, :with_note, :cocina_type, :mods_type
+
+          def types_for(type)
+            return ['uri', 'uri', IdentifierType::STANDARD_IDENTIFIER_SCHEMES] if type == 'uri'
+
+            IdentifierType.cocina_type_for_mods_type(type)
+          end
+
+          def build_source
+            {
+              uri: identifier_element['typeURI']
+            }.tap do |props|
+              props[:code] = mods_type unless props[:uri]
+            end.compact.presence
+          end
+        end
+      end
+    end
+  end
+end

data/lib/cocina/models/mapping/from_mods/identifier_type.rb

@@ -0,0 +1,292 @@
+# frozen_string_literal: true
+
+module Cocina
+  module Models
+    module Mapping
+      module FromMods
+        # Maps identifier types
+        class IdentifierType # rubocop:disable Metrics/ClassLength
+          COCINA_TO_STANDARD_IDENTIFIER_SCHEMES = {
+            'AGROVOC ID' => 'agrovoc',
+            'AllMovie ID' => 'allmovie',
+            'AllMusic ID' => 'allmusic',
+            'AlloCiné ID' => 'allocine',
+            'American National Biography Online ID' => 'amnbo',
+            'ANSI' => 'ansi',
+            'Apis ID' => 'apis',
+            'Artsy ID' => 'artsy',
+            'BFI ID' => 'bfi',
+            'Biblioteca Nacional de España ID' => 'datoses',
+            'Biographical Directory of the United States Congress ID' => 'bdusc',
+            'BnF catalogue général ID' => 'bnfcg',
+            'CANTIC ID' => 'cantic',
+            'CGNDB ID' => 'cgndb',
+            'Danacode ID' => 'danacode',
+            'Det Danske Filminstitut Filmdatabasen ID' => 'dkfilm',
+            'Discogs ID' => 'discogs',
+            'DOI' => 'doi',
+            'EAN' => 'ean',
+            'EIDR ID' => 'eidr',
+            'FAST' => 'fast',
+            'Filmportal ID' => 'filmport',
+            'Find a Grave ID' => 'findagr',
+            'Freebase ID' => 'freebase',
+            'Geographic Names Database ID' => 'geogndb',
+            'GeoNames ID' => 'geonames',
+            'GND ID' => 'gnd',
+            'GNIS ID' => 'gnis',
+            'Gran enciclopèdia catalana ID' => 'gec',
+            'GTIN-14 ID' => 'gtin-14',
+            'Handle' => 'hdl',
+            'IBDB ID' => 'ibdb',
+            'IdRef' => 'idref',
+            'IMDb ID' => 'imdb',
+            'ISAN' => 'isan',
+            'ISBN' => 'isbn',
+            'ISBN registrant element ID' => 'isbnre',
+            'ISIL' => 'isil',
+            'ISMN' => 'ismn',
+            'ISNI' => 'isni',
+            'ISO' => 'iso',
+            'ISRC' => 'isrc',
+            'ISSN' => 'issn',
+            'ISSN-L' => 'issn-l',
+            'ISTC' => 'istc',
+            'ISWC' => 'iswc',
+            'ITAR' => 'itar',
+            'KinoPoisk ID' => 'kinipo',
+            'LC Manuscript Division ID' => 'lcmd',
+            'LCCN' => 'lccn',
+            'Libraries Australia ID' => 'libaus',
+            'local' => 'local',
+            'matrix number' => 'matrix-number',
+            'MOMA ID' => 'moma',
+            'Munzinger ID' => 'munzing',
+            'music plate' => 'music-plate',
+            'music publisher' => 'music-publisher',
+            'MusicBrainz ID' => 'musicb',
+            'Número de Identificación de las Publicaciones Oficiales ID' => 'nipo',
+            'National Gallery of Art ID' => 'nga',
+            'National Portrait Gallery ID' => 'npg',
+            'NNDB ID' => 'nndb',
+            'ODNB ID' => 'odnb',
+            'OpenStreetMap ID' => 'opensm',
+            'ORCID' => 'orcid',
+            'Oxford DNB Index ID' => 'oxforddnb',
+            'PORT.hu ID' => 'porthu',
+            'RBMS Binding Term' => 'rbmsbt',
+            'RBMS Genre Terms ID' => 'rbmsgt',
+            'RBMS Paper Term' => 'rbmspt',
+            'RBMS Printing and Publishing Term' => 'rbmsppe',
+            'RBMS Provenance Evidence' => 'rbmspe',
+            'RBMS Relationship Designator' => 'rbmsrd',
+            'RBMS Type Evidence' => 'rbmste',
+            'ResearcherID' => 'rid',
+            'RKDartists ID' => 'rkda',
+            'Scholar Universe ID' => 'scholaru',
+            'Scope ID' => 'scope',
+            'Scopus Author ID' => 'scopus',
+            'SICI' => 'sici',
+            'Smithsonian American Art Museum ID' => 'saam',
+            'sound recording issue number' => 'issue-number',
+            'Sports Reference: Baseball ID' => 'sprfbsb',
+            'Sports Reference: Basketball ID' => 'sprfbsk',
+            'Sports Reference: College Basketball ID' => 'sprfcbb',
+            'Sports Reference: College Football ID' => 'sprfcfb',
+            'Sports Reference: Hockey ID' => 'sprfhoc',
+            'Sports Reference: Olympic Sports ID' => 'sprfoly',
+            'Sports Reference: Pro Football ID' => 'sprfpfb',
+            'Spotify ID' => 'spotify',
+            'Standard Technical Report Number' => 'strn',
+            'stock number' => 'stock-number',
+            'Svensk Filmdatabas ID' => 'svfilm',
+            'Swets (Netherlands) ID' => 'swets',
+            'Tate Artist ID' => 'tatearid',
+            'TGN' => 'gettytgn',
+            'Theatricalia ID' => 'theatr',
+            'Trove ID' => 'trove',
+            'U.S. National Gazetteer Feature Name ID' => 'natgazfid',
+            'ULAN' => 'gettyulan',
+            'UPC' => 'upc',
+            'URI' => 'uri',
+            'URN' => 'urn',
+            'VIAF' => 'viaf',
+            'video recording number' => 'videorecording-identifier',
+            'Web NDL Authority' => 'wndla',
+            'Wikidata' => 'wikidata'
+          }.freeze
+
+          COCINA_TO_STANDARD_IDENTIFIER_SOURCE_CODES = {
+            'AAT' => 'gettyaat',
+            'actionable ISBN' => 'isbn-a',
+            'AGORHA ID' => 'agorha',
+            'Agricultural Thesaurus and Glossary' => 'atg',
+            'archINFORM location ID' => 'archinl',
+            'archINFORM person ID' => 'archinpe',
+            'archINFORM project ID' => 'archinpr',
+            'Archnet authority' => 'archna',
+            'Archnet site ID' => 'archns',
+            'ARK' => 'ark',
+            'ARK ID' => 'arkid',
+            'Art UK Artists ID' => 'artukart',
+            'Art UK Artworks ID' => 'artukaw',
+            'BALaT People & Institutions ID' => 'balat',
+            'BBC Things ID' => 'bbcth',
+            'Belvedere Artist ID' => 'belvku',
+            'Belvedere Work ID' => 'belvwrk',
+            'Benezit Dictionary of Artists ID' => 'benezit',
+            'BIBBI authority' => 'bibbi',
+            'Biographies of the Entomologists of the World ID' => 'bew',
+            'Biography portal of the Netherlands ID' => 'bpn',
+            'British Standards Institution ID' => 'bsi',
+            'CABI Thesaurus ID' => 'cabt',
+            'Canadiana Authority' => 'cana',
+            'CERL Thesaurus term' => 'cerl',
+            'Cesko-Slovenská filmová databáze ID' => 'csfdcz',
+            'Clara: Database of Women Artists ID' => 'clara',
+            'Collective Biographies of Women Persons ID' => 'cbwpid',
+            'Currículo Lattes ID' => 'lattes',
+            'Dictionnaire des peintres belges ID' => 'dpb',
+            'Digital atlas of the Roman Empire ID' => 'darome',
+            'Early Modern Letters Online ID' => 'emlo',
+            'European Case Law Identifier ID' => 'ecli',
+            'Fide Chess Profile ID' => 'fidecp',
+            'Film Affinity ID' => 'filmaff',
+            'FIS Athlete ID' => 'fisa',
+            'Gemeenschappelijke Thesaurus Audiovisuele Archieven ID' => 'gtaa',
+            'Global Agricultural Concept Scheme ID' => 'gacsch',
+            'Goodreads Author ID' => 'goodra',
+            'Great Britain National Archives ID' => 'nagb',
+            'Great Russian Encyclopedia ID' => 'bigenc',
+            'IAAF Athletes ID' => 'iaafa',
+            'Iconography Authority' => 'iconauth',
+            'Identificativo SBN (Servizio bibliotecario nazionale)' => 'isbnsbn',
+            'ISFDB author directory ID' => 'isfdbau',
+            'ISFDB award directory ID' => 'isfdbaw',
+            'ISFDB magazine directory ID' => 'isfdbma',
+            'ISFDB publisher directory ID' => 'isfdbpu',
+            'J. Paul Getty Museum Artist ID' => 'gettyart',
+            'J. Paul Getty Museum Object ID' => 'gettyobj',
+            'Kunstindeks Danmark Artist ID' => 'kda',
+            'Kunstindeks Danmark Work ID' => 'kdw',
+            'Legal entity identifier' => 'lei',
+            'Marine Gazetteer ID' => 'margaz',
+            'MESH' => 'mesh',
+            'MovieMeter Film ID' => 'moviemetf',
+            'MovieMeter Regisseur ID' => 'moviemetr',
+            "Musée d'Orsay Catalogue des oeuvres fiche oeuvre" => 'mocofo',
+            "Musée d'Orsay Répertoire des artistes notice artiste" => 'morana',
+            'Music Sales Classical ID' => 'muscl',
+            'National Gallery of Victoria Artist ID' => 'ngva',
+            'National Gallery of Victoria Work ID' => 'ngvw',
+            'OFDb ID' => 'ofdb',
+            'ONIX ID' => 'onix',
+            'Pacific Coast Architecture Database building ID' => 'pcadbu',
+            'Pacific Coast Architecture Database firm ID' => 'pcadpf',
+            'Pacific Coast Architecture Database person ID' => 'pcadpe',
+            'PermID' => 'permid',
+            'Personen uit de Nederlandse Thesaurus van Auteursnamen ID' => 'pnta',
+            'PIC ID' => 'picnypl',
+            'Pleiades ID' => 'pleiades',
+            'Prabook ID' => 'prabook',
+            'Quan Guo Bao Kan Suo Yin ID' => 'cnbksy',
+            'ROR' => 'ror',
+            'Russian National Heritage Registry for Books ID' => 'knpam',
+            'S2A3 Biographical Database of Southern African Science ID' => 's2a3bd',
+            'Science Museum Group People ID' => 'smgp',
+            'Semantic Scholar author ID' => 'ssaut',
+            'SNAC' => 'snac',
+            'STW thesaurus ID' => 'stw',
+            'U.S. National Archives Catalog ID' => 'nacat',
+            'UNESCO thesaurus ID' => 'unescot',
+            'Verzeichnis der Drucke des 16. Jahrhunderts ID' => 'vd16',
+            'Verzeichnis der Drucke des 17. Jahrhunderts ID' => 'vd17',
+            'Verzeichnis der Drucke des 18. Jahrhunderts ID' => 'vd18',
+            'VGMdb artist ID' => 'vgmdb',
+            'X Games Athlete ID' => 'xgamea',
+            'ZooBank author ID' => 'zbaut'
+          }.freeze
+
+          OTHER_COCINA_TYPES = [
+            'arXiv',
+            'audio issue number',
+            'audio take',
+            'award number',
+            'barcode',
+            'BIBCODE',
+            'CODEN',
+            'copyright number',
+            'CrossRef Funder ID',
+            'EAN13',
+            'EISSN',
+            'ETD ID',
+            'fingerprint',
+            'GRID',
+            'IGSN',
+            'LC Overseas Acquisition ID',
+            'LISID',
+            'LISSN',
+            'music distributor number',
+            'music opus number',
+            'music serial number',
+            'music thematic number',
+            'NBN',
+            'OCLC',
+            'other system control number',
+            'PMID',
+            'postal registration',
+            'publisher number',
+            'PURL',
+            'report number',
+            'SIRSI',
+            'STM',
+            'study number',
+            'Symphony authority file identifier',
+            'Symphony catkey',
+            'URL',
+            'W3 ID'
+          ].freeze
+
+          ALL_COCINA_TYPES = OTHER_COCINA_TYPES + COCINA_TO_STANDARD_IDENTIFIER_SOURCE_CODES.keys + COCINA_TO_STANDARD_IDENTIFIER_SCHEMES.keys
+
+          STANDARD_IDENTIFIER_SOURCE_CODES = 'standard identifier source codes'
+          STANDARD_IDENTIFIER_SCHEMES = 'standard identifier schemes'
+          COCINA = 'cocina'
+
+          # @param [String] mods_type
+          # @return [String, String, String] cocina type or nil, matched mods type (with correct case) or nil, identifier source or nil
+          def self.cocina_type_for_mods_type(mods_type)
+            return [nil, nil, nil] if mods_type.blank?
+
+            # Try to find a standard identifier scheme type (case insensitive)
+            COCINA_TO_STANDARD_IDENTIFIER_SCHEMES.each_pair do |cocina_type, check_mods_type|
+              if check_mods_type&.downcase == mods_type.downcase
+                return [cocina_type, check_mods_type,
+                        STANDARD_IDENTIFIER_SCHEMES]
+              end
+            end
+
+            # Try to find a standard identifier scheme type (case insensitive)
+            COCINA_TO_STANDARD_IDENTIFIER_SOURCE_CODES.each_pair do |cocina_type, check_mods_type|
+              if check_mods_type&.downcase == mods_type.downcase
+                return [cocina_type, check_mods_type,
+                        STANDARD_IDENTIFIER_SOURCE_CODES]
+              end
+            end
+
+            # Try to find a Cocina type (case insensitive)
+            ALL_COCINA_TYPES.each do |cocina_type|
+              return [cocina_type, nil, COCINA] if cocina_type.casecmp(mods_type).zero?
+            end
+
+            [mods_type, nil, nil]
+          end
+
+          def self.mods_type_for_cocina_type(cocina_type)
+            COCINA_TO_STANDARD_IDENTIFIER_SCHEMES[cocina_type] || COCINA_TO_STANDARD_IDENTIFIER_SOURCE_CODES[cocina_type] || cocina_type
+          end
+        end
+      end
+    end
+  end
+end

data/lib/cocina/models/mapping/from_mods/language.rb

@@ -0,0 +1,36 @@
+# frozen_string_literal: true
+
+module Cocina
+  module Models
+    module Mapping
+      module FromMods
+        # Maps languages
+        class Language
+          # @param [Nokogiri::XML::Element] resource_element mods or relatedItem element
+          # @param [Cocina::Models::Mapping::FromMods::DescriptionBuilder] description_builder
+          # @param [String] purl
+          # @return [Hash] a hash that can be mapped to a cocina model
+          def self.build(resource_element:, description_builder:, purl: nil)
+            new(resource_element: resource_element, description_builder: description_builder).build
+          end
+
+          def initialize(resource_element:, description_builder:)
+            @resource_element = resource_element
+            @notifier = description_builder.notifier
+          end
+
+          def build
+            languages = resource_element.xpath('mods:language', mods: Description::DESC_METADATA_NS).map do |lang_node|
+              Cocina::Models::Mapping::FromMods::LanguageTerm.build(language_element: lang_node, notifier: notifier)
+            end
+            Primary.adjust(languages, 'language', notifier)
+          end
+
+          private
+
+          attr_reader :resource_element, :notifier
+        end
+      end
+    end
+  end
+end

data/lib/cocina/models/mapping/from_mods/language_script.rb

@@ -0,0 +1,30 @@
+# frozen_string_literal: true
+
+module Cocina
+  module Models
+    module Mapping
+      module FromMods
+        # Maps lang and script attributes
+        class LanguageScript
+          # @param [Nokogiri::XML::Element] element that may have lang or script attributes
+          # @param [Cocina::Models::Mapping::FromMods::DescriptionBuilder] description_builder
+          # @return [Hash] a hash that can be mapped to a cocina model for a valueLanguage
+          def self.build(node:)
+            return nil unless node['lang'].present? || node['script'].present?
+
+            {}.tap do |value_language|
+              if node['lang'].present?
+                value_language[:code] = node['lang']
+                value_language[:source] = { code: 'iso639-2b' }
+              end
+              if node['script'].present?
+                value_language[:valueScript] =
+                  { code: node['script'], source: { code: 'iso15924' } }
+              end
+            end
+          end
+        end
+      end
+    end
+  end
+end

data/lib/cocina/models/mapping/from_mods/language_term.rb

@@ -0,0 +1,106 @@
+# frozen_string_literal: true
+
+module Cocina
+  module Models
+    module Mapping
+      module FromMods
+        # Maps language term attributes
+        class LanguageTerm
+          # @param [Nokogiri::XML::Element] language_element language or languageOfCataloging element
+          # @param [Cocina::Models::Mapping::ErrorNotifier] notifier
+          # @return [Hash] a hash that can be mapped to a cocina model
+          def self.build(language_element:, notifier:)
+            new(language_element: language_element, notifier: notifier).build
+          end
+
+          def initialize(language_element:, notifier:)
+            @language_element = language_element
+            @notifier = notifier
+          end
+
+          def build
+            attribs = lang_term_attributes
+            attribs[:status] = status
+            attribs[:script] = script_term_attributes
+            attribs.compact
+          end
+
+          private
+
+          attr_reader :language_element, :notifier
+
+          def lang_term_attributes
+            code_language_term = language_element.xpath('./mods:languageTerm[@type="code"]',
+                                                        mods: Description::DESC_METADATA_NS).first
+            text_language_term = language_element.xpath('./mods:languageTerm[@type="text"]',
+                                                        mods: Description::DESC_METADATA_NS).first
+            if code_language_term.nil? && text_language_term.nil?
+              notifier.warn('languageTerm missing type')
+              code_language_term = language_element.xpath('./mods:languageTerm', mods: Description::DESC_METADATA_NS).first
+            end
+
+            {
+              code: code_language_term&.text,
+              value: text_language_term&.text,
+              uri: ValueURI.sniff(language_value_uri_for(code_language_term, text_language_term), notifier),
+              appliesTo: language_applies_to,
+              displayLabel: language_element['displayLabel']
+            }.tap do |attrs|
+              source = language_source_for(code_language_term, text_language_term)
+              attrs[:source] = source if source.present?
+            end
+          end
+
+          def script_term_attributes
+            script_term_nodes = language_element.xpath('mods:scriptTerm', mods: Description::DESC_METADATA_NS)
+
+            return if script_term_nodes.blank?
+
+            code, value, authority = nil
+            script_term_nodes.each do |script_term_node|
+              code ||= script_term_node.content if script_term_node['type'] == 'code'
+              value ||= script_term_node.content if script_term_node['type'] == 'text'
+              authority ||= script_term_node['authority']
+            end
+            source = { code: authority } if authority
+            {
+              code: code,
+              value: value,
+              source: source
+            }.compact
+          end
+
+          # this can be present for type text and/or code, but we only want one.
+          def language_value_uri_for(code_language_term, text_language_term)
+            code_language_term&.attribute('valueURI')&.to_s || text_language_term&.attribute('valueURI')&.to_s
+          end
+
+          def language_applies_to
+            value = language_element['objectPart']
+            [value: value] if value.present?
+          end
+
+          def language_source_for(code_language_term, text_language_term)
+            {
+              code: code_language_term&.attribute('authority')&.to_s || text_language_term&.attribute('authority')&.to_s,
+              uri: code_language_term&.attribute('authorityURI')&.to_s || text_language_term&.attribute('authorityURI')&.to_s
+
+            }.compact
+          end
+
+          def status
+            status_value = language_element[:usage] || language_element[:status]
+            return unless status_value
+
+            status_value.downcase.tap do |value|
+              if status_value != value
+                notifier.warn("#{language_element.name} usage attribute not downcased",
+                              { value: language_element[:usage] })
+              end
+            end
+          end
+        end
+      end
+    end
+  end
+end