Fix: add format check to the identifier notation (#223)

* add format check to the identifier notation

* Fix test using the good format of identifier notations

Author: imadbourouche

lib/ontologies_linked_data/models/agents/identifier.rb

@@ -6,7 +6,7 @@ class AgentIdentifier < LinkedData::Models::Base
 
       model :Identifier, namespace: :adms, name_with: lambda {  |i| generate_identifier(i.notation, i.schemaAgency)}
 
-      attribute :notation, namespace: :skos, enforce: %i[existence no_url]
+      attribute :notation, namespace: :skos, enforce: %i[existence no_url notation_format]
       attribute :schemaAgency, namespace: :adms, enforcedValues: IDENTIFIER_SCHEMES.keys, enforce: [:existence]
       attribute :schemeURI, handler: :scheme_uri_infer
       attribute :creator, type: :user, enforce: [:existence]
@@ -31,6 +31,33 @@ def no_url(inst,attr)
         return  notation&.start_with?('http') ? [:no_url, "`notation` must not be a URL"]  : []
       end
 
+      def notation_format(inst, attr)
+        inst.bring([attr, :schemaAgency]) if inst.bring?(attr)
+        notation = inst.send(attr)
+        schema_agency = inst.send(:schemaAgency)
+
+        # Validate notation format depending on schema to not have weird ids
+        case schema_agency
+        when "ROR"
+          unless notation.match?(/^[0-9a-z]{9}$/i) # ROR IDs are 9-char base32
+            return [:notation_format, "`notation` must be compliant with ROR format"]
+          end
+        when "ORCID"
+          unless notation.match?(/^\d{4}-\d{4}-\d{4}-\d{3}[\dX]$/)
+            return [:notation_format, "`notation` must be compliant with ORCID format"]
+          end
+        when "ISNI"
+          unless notation.match?(/^\d{4}\s?\d{4}\s?\d{4}\s?\d{3}[\dX]$/)
+            return [:notation_format, "`notation` must be compliant with ISNI format"]
+          end
+        when "GRID"
+          unless notation.match?(/^grid\.[0-9]+\.[a-f0-9]{1,2}$/i)
+            return [:notation_format, "`notation` must be compliant with GRID format"]
+          end
+        end
+
+      end
+
       def scheme_uri_infer
         self.bring(:schemaAgency) if self.bring?(:schemaAgency)
         IDENTIFIER_SCHEMES[self.schemaAgency.to_sym] if self.schemaAgency

test/models/test_agent.rb

@@ -25,7 +25,7 @@ def test_agent_no_valid
     ]
     @identifiers = [
       LinkedData::Models::AgentIdentifier.new(notation: '000h6jb29', schemaAgency: 'ROR', creator: @@user1),
-      LinkedData::Models::AgentIdentifier.new(notation: '000h6jb29', schemaAgency: 'ORCID', creator: @@user1),
+      LinkedData::Models::AgentIdentifier.new(notation: '0000-0012-1501-8134', schemaAgency: 'ORCID', creator: @@user1),
     ]
 
     @identifiers.each { |i| i.save }
@@ -75,7 +75,7 @@ def test_identifier_no_valid
 
     refute LinkedData::Models::AgentIdentifier.new(notation: '000h6jb29', schemaAgency: 'ROR', creator: @@user1).valid?
 
-    assert LinkedData::Models::AgentIdentifier.new(notation: '000h6jb29', schemaAgency: 'ORCID', creator: @@user1).valid?
+    assert LinkedData::Models::AgentIdentifier.new(notation: '0000-0012-1501-8134', schemaAgency: 'ORCID', creator: @@user1).valid?
     id.delete
   end
 

test/models/test_search.rb

@@ -97,7 +97,7 @@ def test_search_agents
     ]
     @identifiers = [
       LinkedData::Models::AgentIdentifier.new(notation: '000h6jb29', schemaAgency: 'ROR', creator: @@user1),
-      LinkedData::Models::AgentIdentifier.new(notation: '000h6jb29', schemaAgency: 'ORCID', creator: @@user1),
+      LinkedData::Models::AgentIdentifier.new(notation: '0000-0012-1501-8134', schemaAgency: 'ORCID', creator: @@user1),
     ]
 
     @identifiers.each { |i| i.save }