Schema 4.6 (#193)

* New 4.6 xsd files

* Add support for new attributes for schema 4.6

* Update the kernel-4 files and test to check updated changes

* Copied the XML from Schema to bolognese

* Revert "Copied the XML from Schema to bolognese"

This reverts commit 15a47cd.

* Implement writing Schema 4.6 values to Schema.org

* Adds support for Schema 4.6 changes in Citeproc and Schema.org reading/writing

* Support for reading Schema 4.6 values from crossref

* Fix for multiple objects in person_group

* Test for Schema 4.6 values in datacite writer

* Updated bolognese gem version

---------

Co-authored-by: Suzanne Vogt <suzanne.vogt@datacite.org>
Co-authored-by: codycooperross <50597551+codycooperross@users.noreply.github.com>

Gemfile.lock

@@ -1,7 +1,7 @@
 PATH
   remote: .
   specs:
-    bolognese (2.2.0)
+    bolognese (2.3.0)
       activesupport (>= 4.2.5)
       benchmark_methods (~> 0.7)
       bibtex-ruby (>= 5.1.0)

lib/bolognese/metadata_utils.rb

@@ -153,6 +153,7 @@ def citeproc_hsh
         "language" => language,
         "author" => author,
         "contributor" => to_citeproc(contributors),
+        "translator" => contributors ? to_citeproc(contributors.select { |c| c["contributorType"] == "Translator" }) : nil,
         "issued" => get_date(dates, "Issued") ? get_date_parts(get_date(dates, "Issued")) : get_date_parts(publication_year.to_s),
         "submitted" => Array.wrap(dates).find { |d| d["dateType"] == "Submitted" }.to_h.fetch("__content__", nil),
         "abstract" => parse_attributes(descriptions, content: "description", first: true),

lib/bolognese/readers/citeproc_reader.rb

@@ -52,6 +52,12 @@ def read_citeproc(string: nil, **options)
           [{ "nameType" => "Organizational", "name" => ":(unav)" }]
         end
         contributors = get_authors(from_citeproc(Array.wrap(meta.fetch("editor", nil))))
+        translators = get_authors(from_citeproc(Array.wrap(meta.fetch("translator", nil))))
+        translators.each do |translator|
+          translator["contributorType"] = "Translator"
+        end
+        contributors += translators
+
         dates = if date = get_date_from_date_parts(meta.fetch("issued", nil))
                   if Date.edtf(date).present?
                     [{ "date" => date,

lib/bolognese/readers/crossref_reader.rb

@@ -4,6 +4,12 @@ module Bolognese
   module Readers
     module CrossrefReader
       # CrossRef types from https://api.crossref.org/types
+
+      CR_TO_DC_CONTRIBUTOR_TYPES = {
+        "editor" => "Editor",
+        "translator" => "Translator",
+      }
+
       def get_crossref(id: nil, **options)
         return { "string" => nil, "state" => "not_found" } unless id.present?
 
@@ -138,7 +144,7 @@ def read_crossref(string: nil, **options)
 
         state = meta.present? || read_options.present? ? "findable" : "not_found"
 
-        related_identifiers = Array.wrap(crossref_is_part_of(journal_metadata)) + Array.wrap(crossref_references(bibliographic_metadata))
+        related_identifiers = Array.wrap(crossref_is_part_of(journal_metadata)) + Array.wrap(crossref_references(bibliographic_metadata)) + Array.wrap(crossref_has_translation(program_metadata)) + Array.wrap(crossref_is_translation_of(program_metadata))
 
         container = if journal_metadata.present?
                       issn = normalize_issn(journal_metadata.to_h.fetch("issn", nil))
@@ -187,7 +193,7 @@ def read_crossref(string: nil, **options)
           "titles" => titles,
           "identifiers" => identifiers,
           "creators" => crossref_people(bibliographic_metadata, "author"),
-          "contributors" => crossref_people(bibliographic_metadata, "editor"),
+          "contributors" => crossref_people(bibliographic_metadata, "editor") + crossref_people(bibliographic_metadata, "translator"),
           "funding_references" => crossref_funding_reference(program_metadata),
           "publisher" => publisher,
           "container" => container,
@@ -276,13 +282,15 @@ def crossref_people(bibliographic_metadata, contributor_role)
               end
             end.compact
 
+            contributor_type = CR_TO_DC_CONTRIBUTOR_TYPES[a["contributor_role"]]
+
             { "nameType" => "Personal",
               "nameIdentifiers" => name_identifiers,
               "name" => [family_name, given_name].compact.join(", "),
               "givenName" => given_name,
               "familyName" => family_name,
               "affiliation" => affiliation.presence,
-              "contributorType" => contributor_role == "editor" ? "Editor" : nil }.compact
+              "contributorType" => contributor_type }.compact
           else
             { "nameType" => "Organizational",
               "name" => a["name"] || a["__content__"] }
@@ -362,6 +370,32 @@ def crossref_references(bibliographic_metadata)
           end
         end.compact.unwrap
       end
+
+      def crossref_has_translation(program_metadata)
+        refs = program_metadata.dig("related_item") if program_metadata.is_a?(Hash)
+        Array.wrap(refs).select { |a| a["intra_work_relation"]["relationship_type"] == "hasTranslation" }.map do |c|
+          if c["intra_work_relation"]["identifier_type"] == "doi"
+            { "relatedIdentifier" => parse_attributes(c["intra_work_relation"]).downcase,
+              "relationType" => "HasTranslation",
+              "relatedIdentifierType" => "DOI" }.compact
+          else
+            nil
+          end
+        end.compact.unwrap
+      end
+
+      def crossref_is_translation_of(program_metadata)
+        refs = program_metadata.dig("related_item") if program_metadata.is_a?(Hash)
+        Array.wrap(refs).select { |a| a["intra_work_relation"]["relationship_type"] == "isTranslationOf" }.map do |c|
+          if c["intra_work_relation"]["identifier_type"] == "doi"
+            { "relatedIdentifier" => parse_attributes(c["intra_work_relation"]).downcase,
+              "relationType" => "IsTranslationOf",
+              "relatedIdentifierType" => "DOI" }.compact
+          else
+            nil
+          end
+        end.compact.unwrap
+      end
     end
   end
 end

lib/bolognese/readers/schema_org_reader.rb

@@ -10,7 +10,9 @@ module SchemaOrgReader
         "isPartOf" => "IsPartOf",
         "hasPart" => "HasPart",
         "isPredecessor" => "IsPreviousVersionOf",
-        "isSuccessor" => "IsNewVersionOf"
+        "isSuccessor" => "IsNewVersionOf",
+        "workTranslation" => "HasTranslation",
+        "translationOfWork" => "IsTranslationOf"
       }
 
       SO_TO_DC_REVERSE_RELATION_TYPES = {
@@ -74,6 +76,12 @@ def read_schema_org(string: nil, **options)
           creators = get_authors(from_schema_org_creators(Array.wrap(authors)))
         end
         contributors = get_authors(from_schema_org_contributors(Array.wrap(meta.fetch("editor", nil))))
+        translators = get_authors(from_schema_org_contributors(Array.wrap(meta.fetch("translator", nil))))
+        translators.map! do |translator|
+          translator["contributorType"] = "Translator"
+          translator
+        end
+        contributors += translators
 
         publisher = {
           "name" => parse_attributes(meta.fetch("publisher", nil), content: "name", first: true),
@@ -106,7 +114,9 @@ def read_schema_org(string: nil, **options)
           Array.wrap(schema_org_references(meta)) +
           Array.wrap(schema_org_is_referenced_by(meta)) +
           Array.wrap(schema_org_is_supplement_to(meta)) +
-          Array.wrap(schema_org_is_supplemented_by(meta))
+          Array.wrap(schema_org_is_supplemented_by(meta)) +
+          Array.wrap(schema_org_has_translation(meta)) +
+          Array.wrap(schema_org_is_translation_of(meta))
 
         rights_list = Array.wrap(meta.fetch("license", nil)).compact.map do |rl|
           hsh_to_spdx("__content__" => rl["name"], "rightsURI" => rl["id"])
@@ -127,6 +137,7 @@ def read_schema_org(string: nil, **options)
         dates << { "date" => meta.fetch("datePublished"), "dateType" => "Issued" } if Date.edtf(meta.fetch("datePublished", nil)).present?
         dates << { "date" => meta.fetch("dateCreated"), "dateType" => "Created" } if Date.edtf(meta.fetch("dateCreated", nil)).present?
         dates << { "date" => meta.fetch("dateModified"), "dateType" => "Updated" } if Date.edtf(meta.fetch("dateModified", nil)).present?
+        dates << { "date" => meta.fetch("temporalCoverage"), "dateType" => "Coverage" } if Date.edtf(meta.fetch("temporalCoverage", nil)).present?
         publication_year = meta.fetch("datePublished")[0..3] if meta.fetch("datePublished", nil).present?
 
         if meta.fetch("inLanguage", nil).is_a?(String)
@@ -240,6 +251,14 @@ def schema_org_is_supplemented_by(meta)
         schema_org_related_identifier(meta, relation_type: "isBasedOn")
       end
 
+      def schema_org_has_translation(meta)
+        schema_org_related_identifier(meta, relation_type: "workTranslation", )
+      end
+
+      def schema_org_is_translation_of(meta)
+        schema_org_related_identifier(meta, relation_type: "translationOfWork")
+      end
+
     end
   end
 end

lib/bolognese/utils.rb

@@ -78,7 +78,9 @@ class << self
       "Other" => "CreativeWork",
       # not part of DataCite schema, but used internally
       "Periodical" => "Periodical",
-      "DataCatalog" => "DataCatalog"
+      "DataCatalog" => "DataCatalog",
+      "Award" => "Grant",
+      "Project" => "Project"
     }
 
     DC_TO_CP_TRANSLATIONS = {
@@ -825,7 +827,8 @@ def to_schema_org_creators(element)
 
     def to_schema_org_contributors(element)
       element = Array.wrap(element).map do |c|
-        c["affiliation"] = Array.wrap(c["affiliation"]).map do |a|
+        transformed_c = c.dup
+        transformed_c["affiliation"] = Array.wrap(c["affiliation"]).map do |a|
           if a.is_a?(String)
             name = a
             affiliation_identifier = nil
@@ -839,10 +842,10 @@ def to_schema_org_contributors(element)
             "@id" => affiliation_identifier,
             "name" => name }.compact
         end.unwrap
-        c["@type"] = c["nameType"].present? ? c["nameType"][0..-3] : nil
-        c["@id"] = Array.wrap(c["nameIdentifiers"]).first.to_h.fetch("nameIdentifier", nil)
-        c["name"] = c["familyName"].present? ? [c["givenName"], c["familyName"]].join(" ") : c["name"]
-        c.except("nameIdentifiers", "nameType").compact
+        transformed_c["@type"] = c["nameType"].present? ? c["nameType"][0..-3] : nil
+        transformed_c["@id"] = Array.wrap(c["nameIdentifiers"]).first.to_h.fetch("nameIdentifier", nil)
+        transformed_c["name"] = c["familyName"].present? ? [c["givenName"], c["familyName"]].join(" ") : c["name"]
+        transformed_c.except("nameIdentifiers", "nameType").compact
       end.unwrap
     end
 
@@ -1234,7 +1237,9 @@ def get_identifier_type(identifier_type)
         "urn" => "URN",
         "md5" => "md5",
         "minid" => "minid",
-        "dataguid" => "dataguid"
+        "dataguid" => "dataguid",
+        "cstr" => "CSTR",
+        "rrid" => "RRID"
       }
 
       identifierTypes[identifier_type.downcase] || identifier_type

lib/bolognese/version.rb

@@ -1,3 +1,3 @@
 module Bolognese
-  VERSION = "2.2.0"
+  VERSION = "2.3.0"
 end

lib/bolognese/writers/schema_org_writer.rb

@@ -13,6 +13,7 @@ def schema_hsh
           "name" => parse_attributes(titles, content: "title", first: true),
           "author" => to_schema_org_creators(creators),
           "editor" => to_schema_org_contributors(contributors),
+          "translator" => contributors ? to_schema_org_contributors(contributors.select { |c| c["contributorType"] == "Translator" }) : nil,
           "description" => parse_attributes(abstract_description, content: "description", first: true),
           "license" => Array.wrap(rights_list).map { |l| l["rightsUri"] }.compact.unwrap,
           "version" => version_info,
@@ -23,6 +24,7 @@ def schema_hsh
           "dateCreated" => get_date(dates, "Created"),
           "datePublished" => get_date(dates, "Issued") || publication_year,
           "dateModified" => get_date(dates, "Updated"),
+          "temporalCoverage" => get_date(dates, "Coverage"),
           "pageStart" => container.to_h["firstPage"],
           "pageEnd" => container.to_h["lastPage"],
           "spatialCoverage" => to_schema_org_spatial_coverage(geo_locations),
@@ -32,6 +34,8 @@ def schema_hsh
           "predecessor_of" => to_schema_org_relation(related_identifiers: related_identifiers, relation_type: "IsPreviousVersionOf"),
           "successor_of" => to_schema_org_relation(related_identifiers: related_identifiers, relation_type: "IsNewVersionOf"),
           "citation" => to_schema_org_relation(related_identifiers: related_identifiers, relation_type: "References"),
+          "workTranslation" => to_schema_org_relation(related_identifiers: related_identifiers, relation_type: "HasTranslation"),
+          "translationOfWork" => to_schema_org_relation(related_identifiers: related_identifiers, relation_type: "IsTranslationOf"),
           "@reverse" => reverse.presence,
           "contentUrl" => Array.wrap(content_url).unwrap,
           "schemaVersion" => schema_version,

resources/kernel-4.6/include/datacite-contributorType-v4.xsd

@@ -0,0 +1,37 @@
+<?xml version="1.0" encoding="UTF-8"?>
+<!-- Version 1.0 - Created 2011-01-13 - FZ, TIB, Germany
+   2013-05 v3.0: Addition of ID to simpleType element, added values "ResearchGroup" & "Other"
+   2014-08-20 v3.1: Addition of value "DataCurator"
+   2015-05-14 v4.0 dropped value "Funder", use new "funderReference"
+   2024-12-31 v4.6: Addition of value "Translator" -->
+<xs:schema xmlns:xs="http://www.w3.org/2001/XMLSchema" xmlns="http://datacite.org/schema/kernel-4" targetNamespace="http://datacite.org/schema/kernel-4" elementFormDefault="qualified">
+  <xs:simpleType name="contributorType" id="contributorType">
+    <xs:annotation>
+      <xs:documentation>The type of contributor of the resource.</xs:documentation>
+    </xs:annotation>
+    <xs:restriction base="xs:string">
+      <xs:enumeration value="ContactPerson" />
+      <xs:enumeration value="DataCollector" />
+      <xs:enumeration value="DataCurator" />
+      <xs:enumeration value="DataManager" />
+      <xs:enumeration value="Distributor" />
+      <xs:enumeration value="Editor" />
+      <xs:enumeration value="HostingInstitution" />
+      <xs:enumeration value="Other" />
+      <xs:enumeration value="Producer" />
+      <xs:enumeration value="ProjectLeader" />
+      <xs:enumeration value="ProjectManager" />
+      <xs:enumeration value="ProjectMember" />
+      <xs:enumeration value="RegistrationAgency" />
+      <xs:enumeration value="RegistrationAuthority" />
+      <xs:enumeration value="RelatedPerson" />
+      <xs:enumeration value="ResearchGroup" />
+      <xs:enumeration value="RightsHolder" />
+      <xs:enumeration value="Researcher" />
+      <xs:enumeration value="Sponsor" />
+      <xs:enumeration value="Supervisor" />
+      <xs:enumeration value="Translator" />
+      <xs:enumeration value="WorkPackageLeader" />
+    </xs:restriction>
+  </xs:simpleType>
+</xs:schema>

resources/kernel-4.6/include/datacite-dateType-v4.xsd

@@ -0,0 +1,27 @@
+<?xml version="1.0" encoding="UTF-8"?>
+<!-- Version 1.0 - Created 2011-01-13 - FZ, TIB, Germany
+     2013-05 v3.0: Addition of ID to simpleType element; addition of value "Collected"; deleted "StartDate" & "EndDate"
+     2017-10-23 v4.1: Addition of value "Other"
+     2019-02-14 v4.2: Addition of value "Withdrawn"
+     2024-12-31 v4.6: Addition of value "Coverage"-->
+<xs:schema xmlns:xs="http://www.w3.org/2001/XMLSchema" xmlns="http://datacite.org/schema/kernel-4" targetNamespace="http://datacite.org/schema/kernel-4" elementFormDefault="qualified">
+  <xs:simpleType name="dateType" id="dateType">
+    <xs:annotation>
+      <xs:documentation>The type of date. Use RKMS‐ISO8601 standard for depicting date ranges.To indicate the end of an embargo period, use Available. To indicate the start of an embargo period, use Submitted or Accepted, as appropriate.</xs:documentation>
+    </xs:annotation>
+    <xs:restriction base="xs:string">
+      <xs:enumeration value="Accepted" />
+      <xs:enumeration value="Available" />
+      <xs:enumeration value="Collected" />
+      <xs:enumeration value="Copyrighted" />
+      <xs:enumeration value="Coverage" />
+      <xs:enumeration value="Created" />
+      <xs:enumeration value="Issued" />
+      <xs:enumeration value="Other" />
+      <xs:enumeration value="Submitted" />
+      <xs:enumeration value="Updated" />
+      <xs:enumeration value="Valid" />
+      <xs:enumeration value="Withdrawn" />
+    </xs:restriction>
+  </xs:simpleType>
+</xs:schema>

resources/kernel-4.6/include/datacite-descriptionType-v4.xsd

@@ -0,0 +1,19 @@
+<?xml version="1.0" encoding="UTF-8"?>
+<!-- Version 1.0 - Created 2011-01-13 - FZ, TIB, Germany
+     2013-05 v3.0: Addition of ID to simpleType element, addition of value "Methods"
+     2015-02-12 v4.0: Addition of value "TechnicalInfo"-->
+<xs:schema xmlns:xs="http://www.w3.org/2001/XMLSchema" xmlns="http://datacite.org/schema/kernel-4" targetNamespace="http://datacite.org/schema/kernel-4" elementFormDefault="qualified">
+  <xs:simpleType name="descriptionType" id="descriptionType">
+    <xs:annotation>
+      <xs:documentation>The type of the description.</xs:documentation>
+    </xs:annotation>
+    <xs:restriction base="xs:string">
+      <xs:enumeration value="Abstract" />
+      <xs:enumeration value="Methods" />
+      <xs:enumeration value="SeriesInformation" />
+      <xs:enumeration value="TableOfContents" />
+      <xs:enumeration value="TechnicalInfo" />
+      <xs:enumeration value="Other" />
+    </xs:restriction>
+  </xs:simpleType>
+</xs:schema>

resources/kernel-4.6/include/datacite-funderIdentifierType-v4.xsd

@@ -0,0 +1,16 @@
+<?xml version="1.0" encoding="UTF-8"?>
+<!-- Version 1.0 - Created 2016-05-14 -->
+<xs:schema xmlns:xs="http://www.w3.org/2001/XMLSchema" xmlns="http://datacite.org/schema/kernel-4" targetNamespace="http://datacite.org/schema/kernel-4" elementFormDefault="qualified">
+  <xs:simpleType name="funderIdentifierType" id="funderIdentifierType">
+    <xs:annotation>
+      <xs:documentation>The type of the funderIdentifier.</xs:documentation>
+    </xs:annotation>
+    <xs:restriction base="xs:string">
+      <xs:enumeration value="ISNI" />
+      <xs:enumeration value="GRID" />
+      <xs:enumeration value="ROR" />
+      <xs:enumeration value="Crossref Funder ID" />
+      <xs:enumeration value="Other" />
+    </xs:restriction>
+  </xs:simpleType>
+</xs:schema>

resources/kernel-4.6/include/datacite-nameType-v4.xsd

@@ -0,0 +1,10 @@
+<?xml version="1.0" encoding="UTF-8"?>
+<!-- Version 4.1 - Created 2017-10-23 -->
+<xs:schema xmlns:xs="http://www.w3.org/2001/XMLSchema" xmlns="http://datacite.org/schema/kernel-4" targetNamespace="http://datacite.org/schema/kernel-4" elementFormDefault="qualified">
+  <xs:simpleType name="nameType" id="nameType">
+    <xs:restriction base="xs:string">
+      <xs:enumeration value="Organizational" />
+      <xs:enumeration value="Personal" />
+    </xs:restriction>
+  </xs:simpleType>
+</xs:schema>

resources/kernel-4.6/include/datacite-numberType-v4.xsd

@@ -0,0 +1,12 @@
+<?xml version="1.0" encoding="UTF-8"?>
+<!-- Version 4.4 - Created 2021-03-05 -->
+<xs:schema xmlns:xs="http://www.w3.org/2001/XMLSchema" xmlns="http://datacite.org/schema/kernel-4" targetNamespace="http://datacite.org/schema/kernel-4" elementFormDefault="qualified">
+  <xs:simpleType name="numberType" id="numberType">
+    <xs:restriction base="xs:string">
+      <xs:enumeration value="Article" />
+      <xs:enumeration value="Chapter" />
+      <xs:enumeration value="Report" />
+      <xs:enumeration value="Other" />
+    </xs:restriction>
+  </xs:simpleType>
+</xs:schema>