Merge pull request #229 from datacite/raid-orcid-skip-fix

Get client when processing nameIdentifiers and skip pushes if DOI is in a raidRegistry

Author: codycooperross

app/models/base.rb

@@ -241,14 +241,14 @@ def self.get_datacite_xml(id)
     Maremma.from_xml(xml).to_h.fetch("resource", {})
   end
 
-  def self.get_datacite_json(id, include_client: false)
+  def self.get_datacite_json(id)
     doi = doi_from_url(id)
     if doi.blank?
       Rails.logger.error "#{id} is not a valid DOI"
       return {}
     end
 
-    url = ENV["API_URL"] + "/dois/#{doi}?affiliation=true#{include_client ? "&include=client" : ""}"
+    url = ENV["API_URL"] + "/dois/#{doi}?affiliation=true"
     response = Maremma.get(url)
 
     if response.status != 200
@@ -257,9 +257,27 @@ def self.get_datacite_json(id, include_client: false)
     end
 
     attributes = (response.body.dig("data", "attributes") || {}).except("xml")
-    included = response.body.dig("included")
-    
-    included.present? ? attributes.merge("included" => included) : attributes
+    relationships = response.body.dig("data", "relationships") || {}
+
+    attributes.merge("relationships" => relationships)
+  end
+
+  def self.get_client(id)
+    url = ENV["API_URL"] + "/clients/#{id}"
+    response = Maremma.get(url)
+    return {} if response.status != 200
+
+    response.body.dig("data", "attributes") || {}
+  end
+
+  def self.raid_registry_record?(attributes)
+    client_id = attributes.dig("relationships", "client", "data", "id")
+    return false if client_id.blank?
+
+    client = cached_client(client_id)
+    return false if client.blank?
+
+    client.dig("clientType") == "raidRegistry"
   end
 
   def self.get_datacite_metadata(id)

app/models/concerns/cacheable.rb

@@ -44,5 +44,11 @@ def cached_crossref_member_id(id)
         Base.get_crossref_member_id(id)
       end
     end
+
+    def cached_client(id)
+      Rails.cache.fetch("clients/#{id}", expires_in: 1.day) do
+        Base.get_client(id)
+      end
+    end
   end
 end

app/models/name_identifier.rb

@@ -34,7 +34,7 @@ def self.import_one(options = {})
       return message
     end
 
-    attributes = get_datacite_json(doi, include_client: true)
+    attributes = get_datacite_json(doi)
     response = push_item({ "id" => doi, "type" => "dois",
                            "attributes" => attributes })
   end
@@ -72,10 +72,12 @@ def self.push_item(item)
     related_identifiers = Array.wrap(attributes.fetch("relatedIdentifiers",
                                                       nil))
 
-    ## Don't process DOIs with certain relationTypes or DOIs in a client with clientType raidRegistry
+    raid_registry_record = raid_registry_record?(attributes)
+
+    ## Don't process DOIs with certain relationTypes or DOIs in a raidRegistry
     skip_doi = related_identifiers.any? do |related_identifier|
       ["IsIdenticalTo", "IsPartOf", "IsPreviousVersionOf", "IsVersionOf"].include?(related_identifier["relationType"] || "")
-    end || item.dig("attributes", "included", 0, "attributes", "clientType") == "raidRegistry"
+    end || raid_registry_record
 
     creators = attributes.fetch("creators", []).select do |n|
       Array.wrap(n.fetch("nameIdentifiers", nil)).any? do |n|