[pull] master from datacite:master

.github/workflows/update-ror-mappings.yml

@@ -22,6 +22,7 @@ jobs:
       BUCKET: ${{ secrets.ROR_ANALYSIS_S3_BUCKET }}
       S3_FUNDER_KEY: ror_funder_mapping/funder_to_ror.json
       S3_HIERARCHY_KEY: ror_funder_mapping/ror_hierarchy.json
+      S3_COUNTRIES_KEY: ror_funder_mapping/ror_to_countries.json
       LOCAL_DIR: app/resources
 
     steps:
@@ -40,6 +41,7 @@ jobs:
           set -euo pipefail
           aws s3 cp "s3://${BUCKET}/${S3_FUNDER_KEY}" funder_to_ror.json.new
           aws s3 cp "s3://${BUCKET}/${S3_HIERARCHY_KEY}" ror_hierarchy.json.new
+          aws s3 cp "s3://${BUCKET}/${S3_COUNTRIES_KEY}" ror_to_countries.json.new
 
       - name: Compare and update tracked files (semantic JSON)
         id: update
@@ -73,6 +75,7 @@ jobs:
 
           normalize_and_update funder_to_ror.json.new "${LOCAL_DIR}/funder_to_ror.json"
           normalize_and_update ror_hierarchy.json.new "${LOCAL_DIR}/ror_hierarchy.json"
+          normalize_and_update ror_to_countries.json.new "${LOCAL_DIR}/ror_to_countries.json"
 
           echo "changed=${changed}" >> "$GITHUB_OUTPUT"
 

app/controllers/datacite_dois_controller.rb

@@ -142,6 +142,7 @@ def index
           client_type: params[:client_type],
           funded_by: params[:funded_by],
           include_funder_child_organizations: params[:include_funder_child_organizations],
+          affiliation_country: params[:affiliation_country],
         )
     end
 
@@ -338,6 +339,7 @@ def index
                       publisher: params[:publisher],
                       funded_by: params[:funded_by],
                       include_funder_child_organizations: params[:include_funder_child_organizations],
+                      "affiliation-country" => params[:affiliation_country],
                       # The cursor link should be an array of values, but we want to encode it into a single string for the URL
                       "page[cursor]" =>
                         page[:cursor] ? make_cursor(results) : nil,

app/models/concerns/rorable.rb

@@ -13,4 +13,10 @@ def get_ror_parents(ror_id)
     hierarchy = RorReferenceStore.ror_hierarchy(normalized_ror)
     hierarchy&.dig("ancestors") || []
   end
+
+  def get_countries_from_ror(ror_id)
+    normalized_ror = "https://#{ror_from_url(ror_id)}"
+    countries = RorReferenceStore.ror_to_countries(normalized_ror)
+    Array.wrap(countries).map(&:upcase).uniq
+  end
 end

app/models/doi.rb

@@ -274,6 +274,7 @@ def validate_publisher_obj?(doi)
       indexes :related_dmp_organization_id,    type: :keyword
       indexes :funder_rors,                    type: :keyword
       indexes :funder_parent_rors,           type: :keyword
+      indexes :affiliation_countries,          type: :keyword
       indexes :client_id_and_name,             type: :keyword
       indexes :provider_id_and_name,           type: :keyword
       indexes :resource_type_id_and_name,      type: :keyword
@@ -648,6 +649,7 @@ def as_indexed_json(_options = {})
       "related_dmp_organization_id" => related_dmp_organization_and_affiliation_id,
       "funder_rors" => funder_rors,
       "funder_parent_rors" => funder_parent_rors,
+      "affiliation_countries" => affiliation_countries,
       "affiliation_id_and_name" => affiliation_id_and_name,
       "fair_affiliation_id_and_name" => fair_affiliation_id_and_name,
       "media_ids" => media_ids,
@@ -1268,6 +1270,14 @@ def self.query(query, options = {})
       minimum_should_match = 1
     end
 
+    if options[:affiliation_country].present?
+      country_codes = options[:affiliation_country]
+                        .split(",")
+                        .map { |c| c.strip.upcase }
+                        .reject(&:blank?)
+      filter << { terms: { "affiliation_countries" => country_codes } } if country_codes.any?
+    end
+
     must_not << { terms: { agency: ["crossref", "kisti", "medra", "jalc", "istic", "airiti", "cnki", "op"] } } if options[:exclude_registration_agencies]
 
     # ES query can be optionally defined in different ways
@@ -2035,6 +2045,32 @@ def funder_parent_rors
     end
   end
 
+  def affiliation_countries
+    countries = []
+    countries.concat(extract_countries_from_people(creators))
+    countries.concat(extract_countries_from_people(contributors))
+    countries.uniq
+  end
+
+  private
+    def extract_countries_from_people(people)
+      Array.wrap(people).flat_map do |person|
+        next [] unless person.is_a?(Hash)
+
+        Array.wrap(person.fetch("affiliation", [])).flat_map do |affiliation|
+          next [] unless affiliation.is_a?(Hash)
+          next [] unless affiliation.fetch("affiliationIdentifierScheme", nil) == "ROR"
+
+          affiliation_identifier = affiliation.fetch("affiliationIdentifier", nil)
+          next [] if affiliation_identifier.blank?
+
+          get_countries_from_ror(affiliation_identifier)
+        end
+      end
+    end
+
+  public
+
   def prefix
     doi.split("/", 2).first if doi.present?
   end

app/services/ror_reference_store.rb

@@ -94,6 +94,12 @@ def refresh!(mapping)
 
       def download_from_s3(filename)
         bucket     = ENV["ROR_ANALYSIS_S3_BUCKET"]
+
+        if bucket.blank?
+          Rails.logger.warn "[RorReferenceStore] ROR_ANALYSIS_S3_BUCKET not configured – skipping S3 fetch for #{filename}"
+          return nil
+        end
+
         object_key = "#{S3_PREFIX}#{filename}"
 
         client   = Aws::S3::Client.new

openapi.yaml

@@ -638,6 +638,12 @@ paths:
           description: Search creators.affiliation.affiliationIdentifier and contributors.affiliation.affiliationIdentifier for a ROR ID.
           schema:
             type: string
+        - in: query
+          name: affiliation-country
+          description: Filter DOIs by associated country inferred from ROR IDs in creators/contributors affiliations. Use comma-separated ISO 3166-1 alpha-2 country codes.
+          schema:
+            type: string
+          example: US,GB
         - in: query
           name: funded-by
           description: Search fundingReferences.funderIdentifier for a ROR ID. Results also include DOIs containing a Crossref Funder ID in fundingReferences.funderIdentifier corresponding to the ROR ID.

spec/concerns/rorable_spec.rb

@@ -10,6 +10,10 @@
       { "ancestors" => ["https://ror.org/04cw6st05"] }
     )
     allow(RorReferenceStore).to receive(:ror_hierarchy).with("https://ror.org/doi.org/00a0jsq62").and_return(nil)
+    allow(RorReferenceStore).to receive(:ror_to_countries).with("https://ror.org/00k4n6c32").and_return(["US"])
+    allow(RorReferenceStore).to receive(:ror_to_countries).with("https://ror.org/00a0jsq62").and_return(["us"])
+    allow(RorReferenceStore).to receive(:ror_to_countries).with("https://ror.org/nonexistent").and_return(nil)
+    allow(RorReferenceStore).to receive(:ror_to_countries).with("https://ror.org/doi.org/00k4n6c32").and_return(nil)
   end
 
   describe "Crossref Funder ID to ROR mapping" do
@@ -61,4 +65,33 @@
       expect(ancestors).to eq([])
     end
   end
+
+  describe "ROR to country mapping" do
+    let(:doi) { create(:doi) }
+
+    it "maps ROR URL to country codes" do
+      expect(doi.get_countries_from_ror("https://ror.org/00k4n6c32")).to eq(["US"])
+    end
+
+    it "maps incomplete ROR URL to country codes" do
+      expect(doi.get_countries_from_ror("ror.org/00k4n6c32")).to eq(["US"])
+    end
+
+    it "maps ROR suffix to country codes" do
+      expect(doi.get_countries_from_ror("00k4n6c32")).to eq(["US"])
+    end
+
+    it "returns empty array for invalid ROR" do
+      expect(doi.get_countries_from_ror("doi.org/00k4n6c32")).to eq([])
+    end
+
+    it "returns empty array for ROR not in mapping" do
+      expect(doi.get_countries_from_ror("https://ror.org/nonexistent")).to eq([])
+    end
+
+    it "normalizes country codes to uppercase" do
+      # Store returns lowercase "us" — method must upcase it
+      expect(doi.get_countries_from_ror("https://ror.org/00a0jsq62")).to eq(["US"])
+    end
+  end
 end

spec/models/doi_spec.rb

@@ -2366,85 +2366,148 @@
     end
   end
 
+  describe "with affiliation ROR IDs" do
+    let(:datacite_ror) { "https://ror.org/00k4n6c32" }
+    let(:cambridge_ror) { "https://ror.org/04wxnsj81" }
+
+    before do
+      allow(RorReferenceStore).to receive(:ror_to_countries).with(datacite_ror).and_return(["US"])
+      allow(RorReferenceStore).to receive(:ror_to_countries).with(cambridge_ror).and_return(["GB"])
+    end
+
+    let(:doi) do
+      create(
+        :doi,
+        creators: [
+          {
+            "name": "Garza, Kristian",
+            "givenName": "Kristian",
+            "familyName": "Garza",
+            "nameType": "Personal",
+            "affiliation": [
+              {
+                "name": "DataCite",
+                "affiliationIdentifier": datacite_ror,
+                "affiliationIdentifierScheme": "ROR",
+              },
+              {
+                "name": "University of Cambridge",
+                "affiliationIdentifier": cambridge_ror,
+                "affiliationIdentifierScheme": "ROR",
+              },
+            ],
+          },
+        ],
+        contributors: [
+          {
+            "name": "Smith, John",
+            "givenName": "John",
+            "familyName": "Smith",
+            "contributorType": "Editor",
+            "affiliation": [
+              {
+                "name": "DataCite",
+                "affiliationIdentifier": datacite_ror,
+                "affiliationIdentifierScheme": "ROR",
+              },
+            ],
+          },
+        ],
+      )
+    end
+
+    let(:expected_countries) { ["US", "GB"] }
+
+    it "has countries from ROR affiliations in affiliation_countries" do
+      expect(doi.affiliation_countries).to match_array(expected_countries)
+      expect(doi.as_indexed_json["affiliation_countries"]).to match_array(expected_countries)
+    end
+
+    it "deduplicates country codes from multiple affiliations" do
+      countries = doi.affiliation_countries
+      expect(countries).to eq(countries.uniq)
+    end
+  end
+
   describe "enrichable" do
-    describe "#enrichment_field" do
-      let(:doi) { create(:doi, aasm_state: "findable", agency: "datacite") }
+  describe "#enrichment_field" do
+    let(:doi) { create(:doi, aasm_state: "findable", agency: "datacite") }
 
-      it "maps alternatveIdentifiers to alternate_identifiers" do
-        expect(doi.enrichment_field("alternateIdentifiers")).to(eq("alternate_identifiers"))
-      end
+    it "maps alternatveIdentifiers to alternate_identifiers" do
+      expect(doi.enrichment_field("alternateIdentifiers")).to(eq("alternate_identifiers"))
+    end
 
-      it "maps creators to creators" do
-        expect(doi.enrichment_field("creators")).to(eq("creators"))
-      end
+    it "maps creators to creators" do
+      expect(doi.enrichment_field("creators")).to(eq("creators"))
+    end
 
-      it "maps titles to titles" do
-        expect(doi.enrichment_field("titles")).to(eq("titles"))
-      end
+    it "maps titles to titles" do
+      expect(doi.enrichment_field("titles")).to(eq("titles"))
+    end
 
-      it "maps publisher to publisher" do
-        expect(doi.enrichment_field("publisher")).to(eq("publisher"))
-      end
+    it "maps publisher to publisher" do
+      expect(doi.enrichment_field("publisher")).to(eq("publisher"))
+    end
 
-      it "maps publicationYear to publication_year" do
-        expect(doi.enrichment_field("publicationYear")).to(eq("publication_year"))
-      end
+    it "maps publicationYear to publication_year" do
+      expect(doi.enrichment_field("publicationYear")).to(eq("publication_year"))
+    end
 
-      it "maps subjects to subjects" do
-        expect(doi.enrichment_field("subjects")).to(eq("subjects"))
-      end
+    it "maps subjects to subjects" do
+      expect(doi.enrichment_field("subjects")).to(eq("subjects"))
+    end
 
-      it "maps contributors to contributors" do
-        expect(doi.enrichment_field("contributors")).to(eq("contributors"))
-      end
+    it "maps contributors to contributors" do
+      expect(doi.enrichment_field("contributors")).to(eq("contributors"))
+    end
 
-      it "maps dates to dates" do
-        expect(doi.enrichment_field("dates")).to(eq("dates"))
-      end
+    it "maps dates to dates" do
+      expect(doi.enrichment_field("dates")).to(eq("dates"))
+    end
 
-      it "maps language to language" do
-        expect(doi.enrichment_field("language")).to(eq("language"))
-      end
+    it "maps language to language" do
+      expect(doi.enrichment_field("language")).to(eq("language"))
+    end
 
-      it "maps types to types" do
-        expect(doi.enrichment_field("types")).to(eq("types"))
-      end
+    it "maps types to types" do
+      expect(doi.enrichment_field("types")).to(eq("types"))
+    end
 
-      it "maps relatedIdentifiers to related_identifiers" do
-        expect(doi.enrichment_field("relatedIdentifiers")).to(eq("related_identifiers"))
-      end
+    it "maps relatedIdentifiers to related_identifiers" do
+      expect(doi.enrichment_field("relatedIdentifiers")).to(eq("related_identifiers"))
+    end
 
-      it "maps relatedItems to related_items" do
-        expect(doi.enrichment_field("relatedItems")).to(eq("related_items"))
-      end
+    it "maps relatedItems to related_items" do
+      expect(doi.enrichment_field("relatedItems")).to(eq("related_items"))
+    end
 
-      it "maps sizes to sizes" do
-        expect(doi.enrichment_field("sizes")).to(eq("sizes"))
-      end
+    it "maps sizes to sizes" do
+      expect(doi.enrichment_field("sizes")).to(eq("sizes"))
+    end
 
-      it "maps formats to formats" do
-        expect(doi.enrichment_field("formats")).to(eq("formats"))
-      end
+    it "maps formats to formats" do
+      expect(doi.enrichment_field("formats")).to(eq("formats"))
+    end
 
-      it "maps version to version" do
-        expect(doi.enrichment_field("version")).to(eq("version"))
-      end
+    it "maps version to version" do
+      expect(doi.enrichment_field("version")).to(eq("version"))
+    end
 
-      it "maps rightsList to rights_list" do
-        expect(doi.enrichment_field("rightsList")).to(eq("rights_list"))
-      end
+    it "maps rightsList to rights_list" do
+      expect(doi.enrichment_field("rightsList")).to(eq("rights_list"))
+    end
 
-      it "maps descriptions to descriptions" do
-        expect(doi.enrichment_field("descriptions")).to(eq("descriptions"))
-      end
+    it "maps descriptions to descriptions" do
+      expect(doi.enrichment_field("descriptions")).to(eq("descriptions"))
+    end
 
-      it "maps geoLocations to geo_locations" do
-        expect(doi.enrichment_field("geoLocations")).to(eq("geo_locations"))
-      end
+    it "maps geoLocations to geo_locations" do
+      expect(doi.enrichment_field("geoLocations")).to(eq("geo_locations"))
+    end
 
-      it "maps fundingReferences to funding_references" do
-        expect(doi.enrichment_field("fundingReferences")).to(eq("funding_references"))
-      end
+    it "maps fundingReferences to funding_references" do
+      expect(doi.enrichment_field("fundingReferences")).to(eq("funding_references"))
     end
   end
 end
+end