Add affiliation-country filter for DOI queries via ROR-to-country mapping

.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

@@ -12,4 +12,12 @@ def get_ror_parents(ror_id)
     normalized_ror = "https://#{ror_from_url(ror_id)}"
     ROR_HIERARCHY[normalized_ror]&.fetch("ancestors", []) || []
   end
+
+  def get_countries_from_ror(ror_id)
+    normalized_ror = ror_from_url(ror_id)
+    return [] if normalized_ror.blank?
+
+    countries = ROR_TO_COUNTRIES[normalized_ror]
+    Array.wrap(countries).map(&:upcase).uniq
+  end
 end

app/models/doi.rb

@@ -272,6 +272,7 @@ 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
@@ -644,6 +645,7 @@ 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,
@@ -1258,6 +1260,14 @@ 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
@@ -2025,6 +2035,33 @@ 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/resources/ror_to_countries.json

@@ -0,0 +1,6 @@
+{
+  "_comment": "Placeholder test data. This file will be populated automatically by the update-ror-mappings workflow from S3.",
+  "ror.org/00k4n6c32": ["US"],
+  "ror.org/00a0jsq62": ["US"],
+  "ror.org/04wxnsj81": ["GB"]
+}

config/initializers/load_ror_data.rb

@@ -2,3 +2,4 @@
 
 FUNDER_TO_ROR = JSON.parse(File.read(Rails.root.join("app/resources/funder_to_ror.json"))).freeze
 ROR_HIERARCHY = JSON.parse(File.read(Rails.root.join("app/resources/ror_hierarchy.json"))).freeze
+ROR_TO_COUNTRIES = JSON.parse(File.read(Rails.root.join("app/resources/ror_to_countries.json"))).freeze

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

@@ -70,4 +70,49 @@
       expect(ancestors).to eq([])
     end
   end
+
+  describe "ROR to country mapping" do
+    let(:doi) { create(:doi) }
+
+    it "loads ROR to countries mapping" do
+      expect(ROR_TO_COUNTRIES).to be_a(Hash)
+      expect(ROR_TO_COUNTRIES).not_to be_empty
+    end
+
+    it "maps ROR URL to country codes" do
+      ror_id = "https://ror.org/00k4n6c32"
+      countries = doi.get_countries_from_ror(ror_id)
+      expect(countries).to eq(["US"])
+    end
+
+    it "maps incomplete ROR URL to country codes" do
+      ror_id = "ror.org/00k4n6c32"
+      countries = doi.get_countries_from_ror(ror_id)
+      expect(countries).to eq(["US"])
+    end
+
+    it "maps ROR suffix to country codes" do
+      ror_id = "00k4n6c32"
+      countries = doi.get_countries_from_ror(ror_id)
+      expect(countries).to eq(["US"])
+    end
+
+    it "returns empty array for invalid ROR" do
+      ror_id = "doi.org/00k4n6c32"
+      countries = doi.get_countries_from_ror(ror_id)
+      expect(countries).to eq([])
+    end
+
+    it "returns empty array for ROR not in mapping" do
+      ror_id = "https://ror.org/nonexistent"
+      countries = doi.get_countries_from_ror(ror_id)
+      expect(countries).to eq([])
+    end
+
+    it "normalizes country codes to uppercase" do
+      ror_id = "https://ror.org/00a0jsq62"
+      countries = doi.get_countries_from_ror(ror_id)
+      expect(countries).to eq(["US"])
+    end
+  end
 end

spec/models/doi_spec.rb

@@ -2363,4 +2363,54 @@
       expect(doi.as_indexed_json["funder_parent_rors"]).to eq(["https://ror.org/019w4f821", "https://ror.org/04cw6st05"])
     end
   end
+
+  describe "with affiliation ROR IDs" do
+    let(:doi) { create(:doi,
+      creators: [
+        {
+          "name": "Garza, Kristian",
+          "givenName": "Kristian",
+          "familyName": "Garza",
+          "nameType": "Personal",
+          "affiliation": [
+            {
+              "name": "DataCite",
+              "affiliationIdentifier": "https://ror.org/00k4n6c32",
+              "affiliationIdentifierScheme": "ROR"
+            },
+            {
+              "name": "University of Cambridge",
+              "affiliationIdentifier": "https://ror.org/04wxnsj81",
+              "affiliationIdentifierScheme": "ROR"
+            }
+          ]
+        }
+      ],
+      contributors: [
+        {
+          "name": "Smith, John",
+          "givenName": "John",
+          "familyName": "Smith",
+          "contributorType": "Editor",
+          "affiliation": [
+            {
+              "name": "DataCite",
+              "affiliationIdentifier": "https://ror.org/00k4n6c32",
+              "affiliationIdentifierScheme": "ROR"
+            }
+          ]
+        }
+      ]
+    ) }
+
+    it "has countries from ROR affiliations in affiliation_countries" do
+      expect(doi.affiliation_countries).to match_array(["US", "GB"])
+      expect(doi.as_indexed_json["affiliation_countries"]).to match_array(["US", "GB"])
+    end
+
+    it "deduplicates country codes from multiple affiliations" do
+      # Verify that the duplicate US from creator and contributor is deduplicated
+      expect(doi.affiliation_countries.count("US")).to eq(1)
+    end
+  end
 end