datacite · svogt0511 · Oct 30, 2025 · May 21, 2025 · May 21, 2025 · May 21, 2025
diff --git a/Gemfile b/Gemfile
@@ -4,7 +4,7 @@ source "https://rubygems.org"
 
 gem "aasm", "~> 5.0", ">= 5.0.1"
 gem "active_model_serializers", "~> 0.10.0"
-gem "activerecord_json_validator", "~> 2.1", ">= 2.1.5"
+gem "activerecord_json_validator", "~> 3.1"
 gem "apollo-federation", "~> 3.8"
 gem "audited", "~> 5.8"
 gem "aws-sdk-core", "~> 3.226"

diff --git a/Gemfile.lock b/Gemfile.lock
@@ -63,9 +63,9 @@ GEM
       activemodel (= 7.2.3)
       activesupport (= 7.2.3)
       timeout (>= 0.4.0)
-    activerecord_json_validator (2.1.5)
-      activerecord (>= 4.2.0, < 8)
-      json_schemer (~> 0.2.18)
+    activerecord_json_validator (3.1.0)
+      activerecord (>= 4.2.0, < 9)
+      json_schemer (~> 2.2)
     activestorage (7.2.3)
       actionpack (= 7.2.3)
       activejob (= 7.2.3)
@@ -277,8 +277,6 @@ GEM
       scanf (~> 1.0)
       sxp (~> 1.2)
       unicode-types (~> 1.7)
-    ecma-re-validator (0.4.0)
-      regexp_parser (~> 2.2)
     edtf (3.2.0)
       activesupport (>= 3.0, < 9.0)
     elasticsearch (7.17.10)
@@ -390,12 +388,11 @@ GEM
     json-ld-preloaded (3.2.2)
       json-ld (~> 3.2)
       rdf (~> 3.2)
-    json_schemer (0.2.25)
-      ecma-re-validator (~> 0.3)
+    json_schemer (2.5.0)
+      bigdecimal
       hana (~> 1.3)
       regexp_parser (~> 2.0)
       simpleidn (~> 0.2)
-      uri_template (~> 0.7)
     jsonapi-renderer (0.2.2)
     jsonapi-serializer (2.2.0)
       activesupport (>= 4.2)
@@ -780,7 +777,6 @@ GEM
     unicode_utils (1.4.0)
     uniform_notifier (1.16.0)
     uri (1.1.1)
-    uri_template (0.7.0)
     useragent (0.16.11)
     uuid (2.3.9)
       macaddr (~> 1.0)
@@ -810,7 +806,7 @@ PLATFORMS
 DEPENDENCIES
   aasm (~> 5.0, >= 5.0.1)
   active_model_serializers (~> 0.10.0)
-  activerecord_json_validator (~> 2.1, >= 2.1.5)
+  activerecord_json_validator (~> 3.1)
   apollo-federation (~> 3.8)
   audited (~> 5.8)
   aws-sdk-core (~> 3.226)

diff --git a/app/models/doi.rb b/app/models/doi.rb
@@ -4,6 +4,14 @@
 require "benchmark"
 
 class Doi < ApplicationRecord
+  INVALID_SCHEMAS = %w[
+      http://datacite.org/schema/kernel-2.1
+      http://datacite.org/schema/kernel-2.2
+      http://datacite.org/schema/kernel-3.0
+      http://datacite.org/schema/kernel-3.1
+      http://datacite.org/schema/kernel-3
+    ].freeze
+
   self.ignored_columns += [:publisher]
   PUBLISHER_JSON_SCHEMA = Rails.root.join("app", "models", "schemas", "doi", "publisher.json")
   audited only: %i[doi url creators contributors titles publisher_obj publication_year types descriptions container sizes formats version_info language dates identifiers related_identifiers related_items funding_references geo_locations rights_list subjects schema_version content_url landing_page aasm_state source reason]
@@ -112,16 +120,13 @@ class Doi < ApplicationRecord
   validates_presence_of :doi
   validates_presence_of :url, if: Proc.new { |doi| doi.is_registered_or_findable? }
 
-  json_schema_validation = {
-    message: ->(errors) { errors },
-    schema: PUBLISHER_JSON_SCHEMA
-  }
-
-  def validate_publisher_obj?(doi)
-    doi.validatable? && doi.publisher_obj? && !(doi.publisher_obj.blank? || doi.publisher_obj.all?(nil))
+  def validate_json_attribute?(attribute)
+    validatable? && !self[attribute].nil? && !INVALID_SCHEMAS.include?(self.schema_version)
   end
 
-  validates :publisher_obj, if: ->(doi) { validate_publisher_obj?(doi) }, json: json_schema_validation
+  def schema_file_path(schema_name)
+    Rails.root.join("app", "models", "schemas", "doi", "#{schema_name}.json")
+  end
 
   # from https://www.crossref.org/blog/dois-and-matching-regular-expressions/ but using uppercase
   validates_format_of :doi, with: /\A10\.\d{4,5}\/[-._;()\/:a-zA-Z0-9*~$=]+\z/, on: :create
@@ -148,6 +153,46 @@ def validate_publisher_obj?(doi)
   validate :check_geo_locations, if: :geo_locations?
   validate :check_language, if: :language?
 
+  # JSON-SCHEMA VALIDATION
+  # temporarily commenting out this validation.
+  validates :identifier, if: proc { |doi| doi.validate_json_attribute?(:identifier) }, json: { message: ->(errors) { errors }, schema: lambda { schema_file_path("identifier") } }, unless: :only_validate
+  validates :creators, if: proc { |doi| doi.validate_json_attribute?(:creators) }, json: { message: ->(errors) { errors }, schema: lambda { schema_file_path("creators") } }, unless: :only_validate
+  validates :titles, if: proc { |doi| doi.validate_json_attribute?(:titles) }, json: { message: ->(errors) { errors }, schema: lambda { schema_file_path("titles") } }, unless: :only_validate
+  validates :publisher_obj, if: proc { |doi| doi.validate_json_attribute?(:publisher_obj) }, json: { message: ->(errors) { errors }, schema: lambda { schema_file_path("publisher") } }, unless: :only_validate
+  validates :publication_year, if: proc { |doi| doi.validate_json_attribute?(:publication_year) }, json: { message: ->(errors) { errors }, schema: lambda { schema_file_path("publication_year") } }, unless: :only_validate
+  validates :subjects, if: proc { |doi| doi.validate_json_attribute?(:subjects) }, json: { message: ->(errors) { errors }, schema: lambda { schema_file_path("subjects") } }, unless: :only_validate
+  validates :contributors, if: proc { |doi| doi.validate_json_attribute?(:contributors) }, json: { message: ->(errors) { errors }, schema: lambda { schema_file_path("contributors") } }, unless: :only_validate
+  validates :dates, if: proc { |doi| doi.validate_json_attribute?(:dates) }, json: { message: ->(errors) { errors }, schema: lambda { schema_file_path("dates") } }, unless: :only_validate
+  validates :alternate_identifiers, if: proc { |doi| doi.validate_json_attribute?(:alternate_identifiers) }, json: { message: ->(errors) { errors }, schema: lambda { schema_file_path("alternate_identifiers") } }, unless: :only_validate
+  validates :related_identifiers, if: proc { |doi| doi.validate_json_attribute?(:related_identifiers) }, json: { message: ->(errors) { errors }, schema: lambda { schema_file_path("related_identifiers") } }, unless: :only_validate
+  validates :sizes, if: proc { |doi| doi.validate_json_attribute?(:sizes) }, json: { message: ->(errors) { errors }, schema: lambda { schema_file_path("sizes") } }, unless: :only_validate
+  validates :formats, if: proc { |doi| doi.validate_json_attribute?(:formats) }, json: { message: ->(errors) { errors }, schema: lambda { schema_file_path("formats") } }, unless: :only_validate
+  validates :version, if: proc { |doi| doi.validate_json_attribute?(:version) }, json: { message: ->(errors) { errors }, schema: lambda { schema_file_path("version") } }, unless: :only_validate
+  validates :rights_list, if: proc { |doi| doi.validate_json_attribute?(:rights_list) }, json: { message: ->(errors) { errors }, schema: lambda { schema_file_path("rights_list") } }, unless: :only_validate
+  validates :descriptions, if: proc { |doi| doi.validate_json_attribute?(:descriptions) }, json: { message: ->(errors) { errors }, schema: lambda { schema_file_path("descriptions") } }, unless: :only_validate
+  validates :geolocations, if: proc { |doi| doi.validate_json_attribute?(:geolocations) }, json: { message: ->(errors) { errors }, schema: lambda { schema_file_path("geolocations") } }, unless: :only_validate
-  validates :geolocations, if: proc { |doi| doi.validate_json_attribute?(:geolocations) }, json: { message: ->(errors) { errors }, schema: lambda { schema_file_path("geolocations") } }, unless: :only_validate
+  validates :geo_locations, if: proc { |doi| doi.validate_json_attribute?(:geo_locations) }, json: { message: ->(errors) { errors }, schema: lambda { schema_file_path("geolocations") } }, unless: :only_validate
-  validates :geolocations, if: proc { |doi| doi.validate_json_attribute?(:geolocations) }, json: { message: ->(errors) { errors }, schema: lambda { schema_file_path("geolocations") } }, unless: :only_validate
+  validates :geo_locations, if: proc { |doi| doi.validate_json_attribute?(:geo_locations) }, json: { message: ->(errors) { errors }, schema: lambda { schema_file_path("geolocations") } }, unless: :only_validate
+  validates :funding_references, if: proc { |doi| doi.validate_json_attribute?(:funding_references) }, json: { message: ->(errors) { errors }, schema: lambda { schema_file_path("funding_references") } }, unless: :only_validate
+  validates :related_items, if: proc { |doi| doi.validate_json_attribute?(:related_items) }, json: { message: ->(errors) { errors }, schema: lambda { schema_file_path("related_items") } }, unless: :only_validate
+
+  validates :raw_language, presence: true, if: proc { |doi| doi.validate_json_attribute?(:raw_language) }, json: {
+    message: ->(errors) { errors },
+    schema: lambda { schema_file_path("language") }
+  }, unless: :only_validate
+
+  validates :raw_types, if: proc { |doi| doi.validate_json_attribute?(:raw_types) }, json: {
+    message: ->(errors) { errors },
+    schema: lambda { schema_file_path("resource_type") },
+  }, unless: :only_validate
+
+  # See https://github.com/mirego/activerecord_json_validator for an explanation of why this must be done.
+  def raw_language
+    self[:language]
+  end
+
+  def raw_types
+    self[:types]
+  end
+
   after_commit :update_url, on: %i[create update]
   after_commit :update_media, on: %i[create update]
 

diff --git a/app/models/schemas/client/subjects.json b/app/models/schemas/client/subjects.json
@@ -1,5 +1,5 @@
 {
-  "$schema": "http://json-schema.org/draft-04/schema#",
+  "$schema": "https://json-schema.org/draft/2020-12/schema",
   "type": "array",
   "minItems": 0,
   "items": {
@@ -14,10 +14,7 @@
           ]
         },
         "lang": {
-          "type": [
-            "string",
-            "null"
-          ]
+          "$ref": "../doi/language.json"
         },
         "subject":             { "type": "string" },
         "subjectScheme":       { "type": "string" }

diff --git a/app/models/schemas/doi/affiliation.json b/app/models/schemas/doi/affiliation.json
@@ -0,0 +1,21 @@
+{
+  "title": "Affiliation",
+  "$schema": "https://json-schema.org/draft/2020-12/schema",
+  "description": "Uniquely identifies an affiliation, according to various identifier schemes.",
+  "type": "object",
+  "properties": {
+    "name": {
+      "$ref": "definitions.json#/$defs/nonemptycontentStringType"
+    },
+    "affiliationIdentifier": {
+      "type": "string"
+    },
+    "affiliationIdentifierScheme": {
+      "type": "string"
+    },
+    "schemeUri": {
+      "type": "string"
+    }
+  },
+  "additionalProperties": false
+}
diff --git a/app/models/schemas/doi/affiliations.json b/app/models/schemas/doi/affiliations.json
@@ -0,0 +1,8 @@
+{
+  "title": "Affiliations",
+  "$schema": "https://json-schema.org/draft/2020-12/schema",
+  "type": "array",
+  "items": {
+    "$ref": "affiliation.json"
+  }
+}
diff --git a/app/models/schemas/doi/alternate_identifier.json b/app/models/schemas/doi/alternate_identifier.json
@@ -0,0 +1,16 @@
+{
+  "title": "AlternateIdentifier",
+  "$schema": "https://json-schema.org/draft/2020-12/schema",
+  "description": "An identifier or identifiers other than the primary Identifier applied to the resource being registered. This may be any alphanumeric string which is unique within its domain of issue. May be used for local identifiers. AlternateIdentifier should be used for another identifier of the same instance (same location, same file).",
+  "type": "object",
+  "properties": {
+    "alternateIdentifier": {
+      "type": "string"
+    },
+    "alternateIdentifierType": {
+      "type": "string"
+    }
+  },
+  "required": ["alternateIdentifierType"],
+  "additionalProperties": false
+}
diff --git a/app/models/schemas/doi/alternate_identifiers.json b/app/models/schemas/doi/alternate_identifiers.json
@@ -0,0 +1,8 @@
+{
+  "title": "AlternateIdentifiers",
+  "$schema": "https://json-schema.org/draft/2020-12/schema",
+  "type": "array",
+  "items": {
+    "$ref": "alternate_identifier.json"
+  }
+}
diff --git a/app/models/schemas/doi/contributor.json b/app/models/schemas/doi/contributor.json
@@ -0,0 +1,16 @@
+{
+  "title": "Contributor",
+  "$schema": "https://json-schema.org/draft/2020-12/schema",
+  "description": "The institution or person responsible for collecting, creating, or otherwise contributing to the development of the dataset.",
+  "type": "object",
+  "$ref": "contributor_base.json",
+  "properties": {
+    "nameIdentifiers": {
+      "$ref": "name_identifiers.json"
+    },
+    "affiliation": {
+      "$ref": "affiliations.json"
+    }
+  },
+  "unevaluatedProperties": false
+}
diff --git a/app/models/schemas/doi/contributor_base.json b/app/models/schemas/doi/contributor_base.json
@@ -0,0 +1,30 @@
+{
+  "title": "ContributorBase",
+  "$schema": "https://json-schema.org/draft/2020-12/schema",
+  "description": "The institution or person responsible for collecting, creating, or otherwise contributing to the development of the dataset.",
+  "type": "object",
+  "properties": {
+    "name": {
+      "$ref": "definitions.json#/$defs/nonemptycontentStringType"
+    },
+    "nameType": {
+      "$ref": "controlled_vocabularies/name_type.json"
+    },
+    "lang": {
+      "$ref": "definitions.json#/$defs/language"
+    },
+    "givenName": {
+      "type": "string"
+    },
+    "familyName": {
+      "type": "string"
+    },
+     "contributorType": {
+      "$ref": "controlled_vocabularies/contributor_type.json"
+    }
+  },
+  "required": [
+    "name",
+    "contributorType"
+  ]
+}
diff --git a/app/models/schemas/doi/contributors.json b/app/models/schemas/doi/contributors.json
@@ -0,0 +1,8 @@
+{
+  "title": "Contributors",
+  "$schema": "https://json-schema.org/draft/2020-12/schema",
+  "type": "array",
+  "items": {
+    "$ref": "contributor.json"
+  }
+}
diff --git a/app/models/schemas/doi/controlled_vocabularies/contributor_type.json b/app/models/schemas/doi/controlled_vocabularies/contributor_type.json
@@ -0,0 +1,30 @@
+{
+  "title": "ContributorType",
+  "$schema": "https://json-schema.org/draft/2020-12/schema",
+  "description": "The type of contributor of the resource.",
+  "type": "string",
+  "enum": [
+    "ContactPerson",
+    "DataCollector",
+    "DataCurator",
+    "DataManager",
+    "Distributor",
+    "Editor",
+    "HostingInstitution",
+    "Producer",
+    "ProjectLeader",
+    "ProjectManager",
+    "ProjectMember",
+    "RegistrationAgency",
+    "RegistrationAuthority",
+    "RelatedPerson",
+    "Researcher",
+    "ResearchGroup",
+    "RightsHolder",
+    "Sponsor",
+    "Supervisor",
+    "Translator",
+    "WorkPackageLeader",
+    "Other"
+  ]
+}
diff --git a/app/models/schemas/doi/controlled_vocabularies/date_type.json b/app/models/schemas/doi/controlled_vocabularies/date_type.json
@@ -0,0 +1,20 @@
+{
+  "title": "DateType",
+  "$schema": "https://json-schema.org/draft/2020-12/schema",
+  "description": "The type of date. Use RKMS‐ISO8601 standard for depicting date ranges.To indicate the end of an embargo period, use Available. To indicate the start of an embargo period, use Submitted or Accepted, as appropriate.",
+  "type": "string",
+  "enum": [
+    "Accepted",
+    "Available",
+    "Copyrighted",
+    "Collected",
+    "Coverage",
+    "Created",
+    "Issued",
+    "Submitted",
+    "Updated",
+    "Valid",
+    "Withdrawn",
+    "Other"
+  ]
+}
diff --git a/app/models/schemas/doi/controlled_vocabularies/description_type.json b/app/models/schemas/doi/controlled_vocabularies/description_type.json
@@ -0,0 +1,14 @@
+{
+  "title": "DescriptionType",
+  "$schema": "https://json-schema.org/draft/2020-12/schema",
+  "description": "The type of description of the resource.",
+  "type": "string",
+  "enum": [
+    "Abstract", 
+    "Methods", 
+    "SeriesInformation", 
+    "TableOfContents", 
+    "TechnicalInfo",
+    "Other"
+  ]
+}
diff --git a/app/models/schemas/doi/controlled_vocabularies/funder_identifier_type.json b/app/models/schemas/doi/controlled_vocabularies/funder_identifier_type.json
@@ -0,0 +1,14 @@
+{
+  "title": "FunderIdentifierType",
+  "$schema": "https://json-schema.org/draft/2020-12/schema",
+  "description": "The type of the funderIdentifier.",
+  "type": "string",
+  "enum": [
+    "ISNI",
+    "GRID",
+    "ROR",
+    "Crossref Funder ID",
+    "Other"
+  ]
+}
+
diff --git a/app/models/schemas/doi/controlled_vocabularies/name_type.json b/app/models/schemas/doi/controlled_vocabularies/name_type.json
@@ -0,0 +1,9 @@
+{
+  "title": "NameType",
+  "$schema": "https://json-schema.org/draft/2020-12/schema",
+  "type": "string",
+  "enum": [
+    "Organizational", 
+    "Personal"
+  ]
+}
diff --git a/app/models/schemas/doi/controlled_vocabularies/number_type.json b/app/models/schemas/doi/controlled_vocabularies/number_type.json
@@ -0,0 +1,11 @@
+{
+  "title": "NumberType",
+  "$schema": "https://json-schema.org/draft/2020-12/schema",
+  "type": "string",
+  "enum": [
+    "Article",
+    "Chapter",
+    "Report",
+    "Other"
+  ]
+}