Merge branch 'master' of https://github.com/datacite/lupo into schema…

…-4.5-r3

app/models/doi.rb

@@ -258,6 +258,7 @@ def validate_publisher_obj?(doi)
       indexes :provider_id,                    type: :keyword
       indexes :consortium_id,                  type: :keyword
       indexes :resource_type_id,               type: :keyword
+      indexes :person_id,                      type: :keyword
       indexes :affiliation_id,                 type: :keyword
       indexes :fair_affiliation_id,            type: :keyword
       indexes :organization_id,                type: :keyword
@@ -563,6 +564,14 @@ def validate_publisher_obj?(doi)
       indexes :fields_of_science, type: :keyword
       indexes :fields_of_science_combined, type: :keyword
       indexes :fields_of_science_repository, type: :keyword
+      indexes :related_doi, type: :object, properties: {
+        client_id: { type: :keyword },
+        doi: { type: :keyword },
+        organization_id: { type: :keyword },
+        person_id: { type: :keyword },
+        resource_type_id: { type: :keyword },
+        resource_type_id_and_name: { type: :keyword },
+      }
       indexes :publisher_obj, type: :object, properties: {
         name: { type: :text, fields: { keyword: { type: "keyword" } } },
         publisherIdentifier: { type: :keyword, normalizer: "keyword_lowercase" },
@@ -1681,6 +1690,10 @@ def consortium_id
     client.provider.consortium_id.downcase if client.present? && client.provider.consortium_id.present?
   end
 
+  def related_dois
+    Doi::Indexer::RelatedDoiIndexer.new(related_identifiers).as_indexed_json
+  end
+
   def related_dmp_ids
     Array.wrap(related_identifiers).select { |related_identifier|
       related_identifier["relatedIdentifierType"] == "DOI"
@@ -1711,6 +1724,19 @@ def sponsor_contributors
   end
 
 
+  def person_id
+    (Array.wrap(creators) + Array.wrap(contributors)).reduce([]) do |sum, c|
+      Array.wrap(c.fetch("nameIdentifiers", nil)).each do |name_identifier|
+        if name_identifier.is_a?(Hash) && name_identifier.fetch("nameIdentifierScheme", nil) == "ORCID" && name_identifier.fetch("nameIdentifier", nil).present?
+          sum << orcid_as_url(
+            orcid_from_url(name_identifier.fetch("nameIdentifier", nil))
+          )
+        end
+      end
+      sum.uniq
+    end
+  end
+
   def organization_id
     organization_ids = (Array.wrap(creators) + Array.wrap(contributors)).reduce([]) do |sum, c|
       Array.wrap(c.fetch("nameIdentifiers", nil)).each do |name_identifier|
@@ -1719,7 +1745,7 @@ def organization_id
       sum
     end
     organization_ids << ror_from_url(publisher["publisherIdentifier"]) if publisher_has_ror?
-    organization_ids
+    organization_ids.uniq
   end
 
   def publisher_has_ror?
@@ -1731,9 +1757,11 @@ def publisher_has_ror?
   def fair_organization_id
     (Array.wrap(creators) + sponsor_contributors).reduce([]) do |sum, c|
       Array.wrap(c.fetch("nameIdentifiers", nil)).each do |name_identifier|
-        sum << ror_from_url(name_identifier.fetch("nameIdentifier", nil)) if name_identifier.is_a?(Hash) && name_identifier.fetch("nameIdentifierScheme", nil) == "ROR" && name_identifier.fetch("nameIdentifier", nil).present?
+        if name_identifier.is_a?(Hash) && name_identifier.fetch("nameIdentifierScheme", nil) == "ROR" && name_identifier.fetch("nameIdentifier", nil).present?
+          sum << ror_from_url(name_identifier.fetch("nameIdentifier", nil))
+        end
       end
-      sum
+      sum.uniq
     end
   end
 
@@ -1742,7 +1770,7 @@ def affiliation_id
       Array.wrap(c.fetch("affiliation", nil)).each do |affiliation|
         sum << ror_from_url(affiliation.fetch("affiliationIdentifier", nil)) if affiliation.is_a?(Hash) && affiliation.fetch("affiliationIdentifierScheme", nil) == "ROR" && affiliation.fetch("affiliationIdentifier", nil).present?
       end
-      sum
+      sum.uniq
     end
   end
 
@@ -1751,7 +1779,7 @@ def fair_affiliation_id
       Array.wrap(c.fetch("affiliation", nil)).each do |affiliation|
         sum << ror_from_url(affiliation.fetch("affiliationIdentifier", nil)) if affiliation.is_a?(Hash) && affiliation.fetch("affiliationIdentifierScheme", nil) == "ROR" && affiliation.fetch("affiliationIdentifier", nil).present?
       end
-      sum
+      sum.uniq
     end
   end
 
@@ -1760,7 +1788,7 @@ def affiliation_id_and_name
       Array.wrap(c.fetch("affiliation", nil)).each do |affiliation|
         sum << "#{ror_from_url(affiliation.fetch('affiliationIdentifier', nil))}:#{affiliation.fetch('name', nil)}" if affiliation.is_a?(Hash) && affiliation.fetch("affiliationIdentifierScheme", nil) == "ROR" && affiliation.fetch("affiliationIdentifier", nil).present?
       end
-      sum
+      sum.uniq
     end
   end
 
@@ -1769,7 +1797,7 @@ def fair_affiliation_id_and_name
       Array.wrap(c.fetch("affiliation", nil)).each do |affiliation|
         sum << "#{ror_from_url(affiliation.fetch('affiliationIdentifier', nil))}:#{affiliation.fetch('name', nil)}" if affiliation.is_a?(Hash) && affiliation.fetch("affiliationIdentifierScheme", nil) == "ROR" && affiliation.fetch("affiliationIdentifier", nil).present?
       end
-      sum
+      sum.uniq
     end
   end
 

app/models/doi/indexer/related_doi_indexer.rb

@@ -0,0 +1,45 @@
+# frozen_string_literal: true
+
+module Doi::Indexer
+  class RelatedDoiIndexer
+    def initialize(related_identifiers)
+      @related_identifiers = related_identifiers
+      @related_dois = nil
+    end
+
+    def related_dois
+      @related_dois ||= @related_identifiers.select { |r| r["relatedIdentifierType"] == "DOI" }
+    end
+
+    def related_grouped_by_id
+      related_dois.group_by { |r| r["relatedIdentifier"].downcase }
+    end
+
+    def relation_types_gouped_by_id
+      related_grouped_by_id.transform_values do |values|
+        values.map { |val| val["relationType"].underscore }.uniq
+      end
+    end
+
+    def related_doi_ids
+      related_grouped_by_id.keys
+    end
+
+    def dois
+      Doi.where(doi: related_doi_ids)
+    end
+
+    def indexed_dois
+      dois.map { |d| RelatedIdentifierDenormalizer.new(d).to_hash }
+    end
+
+    def as_indexed_json
+      rtypes = relation_types_gouped_by_id
+      indexed_dois.map do |indexed_doi|
+        doi = indexed_doi["doi"]
+        indexed_doi["relation_type"] = rtypes[doi]
+        indexed_doi
+      end
+    end
+  end
+end

app/models/doi/indexer/related_identifier_denormalizer.rb

@@ -0,0 +1,32 @@
+# frozen_string_literal: true
+
+module Doi::Indexer
+  class RelatedIdentifierDenormalizer
+    attr_reader :related_doi
+
+    def initialize(doi)
+      @related_doi = doi
+    end
+
+    def to_hash
+      %w[
+        client_id
+        doi
+        organization_id
+        person_id
+        resource_type_id
+        resource_type_id_and_name
+      ].index_with { |method_name| send(method_name) }
+    end
+
+    delegate :resource_type_id, to: :related_doi
+    delegate :resource_type_id_and_name, to: :related_doi
+    delegate :organization_id, to: :related_doi
+    delegate :person_id, to: :related_doi
+    delegate :client_id, to: :related_doi
+
+    def doi
+      @related_doi.doi.downcase
+    end
+  end
+end

spec/models/doi_spec.rb

@@ -1112,6 +1112,66 @@
     end
   end
 
+  describe "person_ids" do
+    it "from creators and contributors" do
+      subject = build(
+        :doi,
+        creators: [
+          {
+            "familyName" => "Garza",
+            "givenName" => "Kristian",
+            "name" => "Garza, Kristian",
+            "nameIdentifiers" => [
+              {
+                "nameIdentifier" => "https://orcid.org/0000-0003-3484-6875",
+                "nameIdentifierScheme" => "ORCID",
+                "schemeUri" => "https://orcid.org",
+              },
+            ],
+            "nameType" => "Personal",
+            "affiliation" => [
+              {
+                "name" => "University of Cambridge",
+                "affiliationIdentifier" => "https://ror.org/013meh722",
+                "affiliationIdentifierScheme" => "ROR",
+              },
+           ],
+          },
+        ],
+        contributors: [
+          {
+            "contributorType" => "Sponsor",
+            "familyName" => "Bob",
+            "givenName" => "Jones",
+            "name" => "Jones, Bob",
+            "nameIdentifiers" => [
+              {
+                "nameIdentifier" => "https://orcid.org/0000-0003-3484-0000",
+                "nameIdentifierScheme" => "ORCID",
+                "schemeUri" => "https://orcid.org",
+              },
+            ],
+            "nameType" => "Personal",
+            "affiliation" => [
+              {
+                "name" => "University of Examples",
+                "affiliationIdentifier" => "https://ror.org/013meh8888",
+                "affiliationIdentifierScheme" => "ROR",
+              },
+           ],
+          },
+        ]
+      )
+      expect(subject).to be_valid
+      expect(subject.person_id).to eq(
+        [
+          "https://orcid.org/0000-0003-3484-6875",
+          "https://orcid.org/0000-0003-3484-0000",
+        ]
+      )
+    end
+  end
+
   describe "related_identifiers" do
     it "has part" do
       subject = build(:doi, related_identifiers: [

spec/models/doi_verified_related_identifiers_spec.rb

@@ -0,0 +1,60 @@
+
+# frozen_string_literal: true
+
+require "rails_helper"
+
+describe Doi, type: :model, vcr: true, elasticsearch: true do
+  describe "related_doi" do
+    let(:client) { create(:client) }
+    let(:target_doi) do
+      create(:doi,
+        client: client,
+        aasm_state: "findable",
+        types: { "resourceTypeGeneral" => "Dataset" }
+      )
+    end
+    let(:doi) do
+      create(:doi,
+        client: client,
+        aasm_state: "findable",
+        related_identifiers: [
+          {
+            "relatedIdentifier": target_doi.doi,
+            "relatedIdentifierType": "DOI",
+            "relationType": "HasPart",
+            "resourceTypeGeneral": "OutputManagementPlan",
+          },
+          {
+            "relatedIdentifier": target_doi.doi,
+            "relatedIdentifierType": "DOI",
+            "relationType": "Cites",
+            "resourceTypeGeneral": "Text",
+          }
+        ])
+    end
+
+    it "indexes related_dois" do
+      expect(doi.related_dois.first["doi"]).to eq(target_doi.doi.downcase)
+    end
+
+    it "indexes related doi's client_id" do
+      expect(doi.related_dois.first["client_id"]).to eq(target_doi.client_id)
+    end
+
+    it "indexes related doi's person_id" do
+      expect(doi.related_dois.first["person_id"]).to eq(target_doi.person_id)
+    end
+
+    it "does not index related doi's claimed resource_type_id" do
+      expect(doi.related_dois.first["resource_type_id"]).not_to eq("output_management_plan")
+    end
+
+    it "indexes related doi's true resource_type_id" do
+      expect(doi.related_dois.first["resource_type_id"]).to eq("dataset")
+    end
+
+    it "indexes all relations to the related doi" do
+      expect(doi.related_dois.first["relation_type"]).to eq(["has_part", "cites"])
+    end
+  end
+end