Enforce data structure and data type consistency for JSON metadata

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", "1.1.3"
 gem "audited", "~> 5.4", ">= 5.4.3"
 gem "aws-sdk-s3"

Gemfile.lock

@@ -66,9 +66,9 @@ GEM
       activemodel (= 7.1.3.2)
       activesupport (= 7.1.3.2)
       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 (7.1.3.2)
       actionpack (= 7.1.3.2)
       activejob (= 7.1.3.2)
@@ -278,8 +278,6 @@ 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 (2.4.0)
+      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)
@@ -771,7 +768,6 @@ GEM
     unicode_utils (1.4.0)
     uniform_notifier (1.16.0)
     uri (0.13.2)
-    uri_template (0.7.0)
     uuid (2.3.9)
       macaddr (~> 1.0)
     uuidtools (2.2.0)
@@ -799,7 +795,7 @@ 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 (= 1.1.3)
   audited (~> 5.4, >= 5.4.3)
   aws-sdk-s3

app/models/doi.rb

@@ -109,16 +109,19 @@ 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))
   end
 
-  validates :publisher_obj, if: ->(doi) { validate_publisher_obj?(doi) }, json: json_schema_validation
+  def validate_json_attribute?(attribute)
+    validatable? && !self[attribute].nil?
+  end
+
+  def schema_file_path(schema_name)
+    Rails.root.join("app", "models", "schemas", "doi", "#{schema_name}.json")
+  end
+
+  validates :publisher_obj, if: ->(doi) { validate_publisher_obj?(doi) }, json: { message: ->(errors) { errors }, schema: lambda { schema_file_path("publisher") } }
 
   # 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
@@ -131,6 +134,7 @@ def validate_publisher_obj?(doi)
   validate :check_dates, if: :dates?
   validate :check_rights_list, if: :rights_list?
   validate :check_titles, if: :titles?
+  validates :titles, if: proc { |doi| doi.validate_json_attribute?(:titles) }, json: { message: ->(errors) { errors }, schema: lambda { schema_file_path("titles") } }
   validate :check_descriptions, if: :descriptions?
   validate :check_types, if: :types?
   validate :check_container, if: :container?

app/models/schemas/doi/controlled_vocabularies/title_type.json

@@ -0,0 +1,13 @@
+{
+  "type": ["string", "null"],
+  "$schema": "http://json-schema.org/draft-04/schema#",
+  "anyOf": [
+    {
+      "type": "string",
+      "enum": ["AlternativeTitle", "Subtitle", "TranslatedTitle", "Other"]
+    },
+    {
+      "type": "null"
+    }
+  ]
+}

app/models/schemas/doi/titles.json

@@ -0,0 +1,19 @@
+{
+  "type": "array",
+  "$schema": "https://json-schema.org/draft/2020-12/schema",
+  "items": {
+    "type": "object",
+    "properties": {
+      "title": { 
+        "type": "string"
+      },
+      "titleType": { "$ref": "controlled_vocabularies/title_type.json" },
+      "lang": { "type": ["string", "null"] }
+    },
+    "required": [
+      "title"
+    ],
+    "additionalProperties": false
+  },
+  "minitems": 1
+}

spec/concerns/indexable_spec.rb

@@ -69,7 +69,7 @@
     let!(:doi) do
       create(
         :doi,
-        titles: { title: "Soil investigations" },
+        titles: [ { title: "Soil investigations" } ],
         publisher: "Pangaea",
         descriptions: { description: "this is a description" },
         aasm_state: "findable",

spec/fixtures/files/nasa_error.json

@@ -31,9 +31,11 @@
           "familyName":"Ramasubramanian"
         }
   		],
-  		"titles":{
+  		"titles": [ 
+        {
   			"title":"Tropical Cyclone Satellite Imagery and Wind Speed Dataset"
-  		},
+        }
+      ],
   		"publisher": "Radiant MLHub",
   		"publicationYear": 2020,
   		"types":{

spec/requests/datacite_dois/patch_spec.rb

@@ -417,7 +417,7 @@
         patch "/dois/#{doi.doi}", valid_attributes, headers
 
         expect(last_response.status).to eq(422)
-        expect(json["errors"]).to eq([{ "source" => "titles", "title" => "Title 'Submitted chemical data for InChIKey=YAPQBXQYLJRXSA-UHFFFAOYSA-N' should be an object instead of a string.", "uid" => "10.14454/4k3m-nyvg" }])
+        expect(json["errors"]).to eq([{ "source" => "titles", "title" => "Value at root is not an array", "uid" => "10.14454/4k3m-nyvg" }])
       end
     end
 

spec/requests/datacite_dois/post_spec.rb

@@ -1274,7 +1274,7 @@
     end
 
     context "when the title changes" do
-      let(:titles) { { "title" => "Referee report. For: RESEARCH-3482 [version 5; referees: 1 approved, 1 approved with reservations]" } }
+      let(:titles) { [ { "title" => "Referee report. For: RESEARCH-3482 [version 5; referees: 1 approved, 1 approved with reservations]" } ] }
       let(:xml) { Base64.strict_encode64(file_fixture("datacite.xml").read) }
       let(:valid_attributes) do
         {