Merge pull request #189 from datacite/codycooperross/issue188

Fix for ORCID and ROR identifiers ending with a `/` not being rendered in JSON when converting from XML

Author: svogt0511

lib/bolognese/utils.rb

@@ -600,12 +600,12 @@ def orcid_as_url(orcid)
     end
 
     def validate_orcid(orcid)
-      orcid = Array(/\A(?:(?:http|https):\/\/(?:(?:www|sandbox)?\.)?orcid\.org\/)?(\d{4}[[:space:]-]\d{4}[[:space:]-]\d{4}[[:space:]-]\d{3}[0-9X]+)\z/.match(orcid)).last
+      orcid = Array(/\A(?:(?:http|https):\/\/(?:(?:www|sandbox)?\.)?orcid\.org\/)?(\d{4}[[:space:]-]\d{4}[[:space:]-]\d{4}[[:space:]-]\d{3}[0-9X]+)\/{0,1}\z/.match(orcid)).last
       orcid.gsub(/[[:space:]]/, "-") if orcid.present?
     end
 
     def validate_ror(ror)
-      Array(/^(?:(?:(?:http|https):\/\/)?ror\.org\/)?(0\w{6}\d{2})$/.match(ror)).last
+      Array(/^(?:(?:(?:http|https):\/\/)?ror\.org\/)?(0\w{6}\d{2})\/{0,1}$/.match(ror)).last
     end
 
     def validate_orcid_scheme(orcid_scheme)

spec/author_utils_spec.rb

@@ -168,11 +168,12 @@
     expect(subject.creators[4]).to eq("nameType"=>"Organizational", "name"=>"University Of Kivu", "nameIdentifiers"=> [{"nameIdentifier"=>"https://ror.org/01qfhxr31", "schemeUri"=>"https://ror.org", "nameIdentifierScheme"=>"ROR"}], "affiliation"=>[])
     expect(subject.creators[5]).to eq("nameType"=>"Organizational", "name"=>"សាកលវិទ្យាល័យកម្ពុជា", "nameIdentifiers"=> [{"nameIdentifier"=>"http://ror.org/025e3rc84", "nameIdentifierScheme"=>"RORS"}], "affiliation"=>[])
     expect(subject.creators[6]).to eq("nameType"=>"Organizational", "name"=>"جامعة زاخۆ", "nameIdentifiers"=> [{"nameIdentifier"=>"05sd1pz50", "schemeUri"=>"https://ror.org", "nameIdentifierScheme"=>"RORS"}], "affiliation"=>[])
+    expect(subject.creators[9]).to eq("nameType"=>"Organizational", "name"=>"Gump South Pacific Research Station", "nameIdentifiers"=> [{"nameIdentifier"=>"https://ror.org/04sk0et52", "schemeUri"=>"https://ror.org", "nameIdentifierScheme"=>"ROR"}], "affiliation"=>[])
     expect(subject.contributors.first).to eq("nameType"=>"Organizational", "name"=>" Nawroz University ", "nameIdentifiers"=> [{"nameIdentifier"=>"https://ror.org/04gp75d48", "schemeUri"=>"https://ror.org", "nameIdentifierScheme"=>"ROR"}], "affiliation"=>[], "contributorType"=>"Producer")
     expect(subject.contributors.last).to eq("nameType"=>"Organizational", "name"=>"University Of Greenland (Https://Www.Uni.Gl/)", "nameIdentifiers"=> [{"nameIdentifier"=>"https://ror.org/00t5j6b61", "schemeUri"=>"https://ror.org", "nameIdentifierScheme"=>"ROR"}],"affiliation"=>[], "contributorType"=>"Sponsor")
   end
 
-  context "affiliationIdentifier" do
+  context "affiliationIdentifier/nameIdentifier" do
     let(:input) { fixture_path + 'datacite-example-ROR-nameIdentifiers.xml' }
     subject { Bolognese::Metadata.new(input: input, from: "datacite") }
 
@@ -207,6 +208,11 @@
       expect(subject.creators[8]["nameIdentifiers"]).to eq([{"nameIdentifier"=>"https://orcid.org/0000-0001-9998-0118", "schemeUri"=>"https://orcid.org", "nameIdentifierScheme"=>"ORCID"}])
     end
 
+    it "should normalize valid ORCID nameIdentifier with trailing slash" do
+      #"  0000-0001-9998-0118  ",  # Valid ORCID with leading/trailing spaces
+      expect(subject.creators[10]["nameIdentifiers"]).to eq([{"nameIdentifier"=>"https://orcid.org/0000-0001-9998-0117", "schemeUri"=>"https://orcid.org", "nameIdentifierScheme"=>"ORCID"}])
+    end
+
     it "should parse non ROR schema's without normalizing them" do
       input = fixture_path + 'datacite-example-ROR-nameIdentifiers.xml'
       subject = Bolognese::Metadata.new(input: input, from: "datacite")
@@ -250,4 +256,4 @@
       expect(response).to eq("Matt Jones and Peter Slaughter and {University of California, Santa Barbara}")
     end
   end
-end
+end

spec/fixtures/datacite-example-ROR-nameIdentifiers.xml

@@ -43,6 +43,14 @@
             <creatorName nameType="Personal">Mike B</creatorName>
             <nameIdentifier nameIdentifierScheme="ORCID"> 0000-0001-9998-0118 </nameIdentifier>
         </creator>
+        <creator>
+            <creatorName nameType="Organizational">Gump South Pacific Research Station</creatorName>
+            <nameIdentifier nameIdentifierScheme="ROR" schemeURI="https://ror.org/">https://ror.org/04sk0et52/</nameIdentifier>
+        </creator>
+        <creator>
+            <creatorName nameType="Personal">Ashwini Sukale</creatorName>
+            <nameIdentifier schemeURI="https://orcid.org/" nameIdentifierScheme="ORCID">https://orcid.org/0000-0001-9998-0117/</nameIdentifier>
+        </creator>
     </creators>
     <titles>
         <title xml:lang="en-US">Genomic Standards Consortium (GSC) Island Sampling Day: Moorea Reef to Ridges Genomic Transect</title>