Fix: multilingual generated missing labels issue (#101)

* make the api use submission naturalLanguage to choose the default language to display * we use only BRO v3.2 in test_classes_controller now, so need of this check
ontoportal-lirmm · Oct 24, 2024 · 34ecc2d · 34ecc2d
1 parent 3fb29ce
commit 34ecc2d
Show file tree

Hide file tree

Showing 4 changed files with 79 additions and 27 deletions.
diff --git a/Gemfile.lock b/Gemfile.lock
@@ -11,7 +11,7 @@ GIT
 
 GIT
   remote: https://github.com/ontoportal-lirmm/goo.git
-  revision: 6018b33373467b778744432ec78a6d814159d129
+  revision: f8ac7b00e8d8b46d1eea04de014175525c1cdd83
   branch: development
   specs:
     goo (0.0.2)
@@ -57,7 +57,7 @@ GIT
 
 GIT
   remote: https://github.com/ontoportal-lirmm/ontologies_linked_data.git
-  revision: 0a456557bc0ae2a7d016000de9b57d3f9eca99a5
+  revision: 793c4fde592e6db91d8d103c15afe993c3472aae
   branch: development
   specs:
     ontologies_linked_data (0.0.1)
@@ -204,7 +204,7 @@ GEM
       grpc (~> 1.41)
     googleapis-common-protos-types (1.16.0)
       google-protobuf (>= 3.18, < 5.a)
-    googleauth (1.11.1)
+    googleauth (1.11.2)
       faraday (>= 1.0, < 3.a)
       google-cloud-env (~> 2.1)
       jwt (>= 1.4, < 3.0)
@@ -225,7 +225,7 @@ GEM
     httpclient (2.8.3)
     i18n (0.9.5)
       concurrent-ruby (~> 1.0)
-    json (2.7.2)
+    json (2.7.3)
     json-ld (3.0.2)
       multi_json (~> 1.12)
       rdf (>= 2.2.8, < 4.0)
@@ -343,7 +343,7 @@ GEM
       mime-types (>= 1.16, < 4.0)
       netrc (~> 0.8)
     retriable (3.1.2)
-    rexml (3.3.8)
+    rexml (3.3.9)
     rsolr (2.6.0)
       builder (>= 2.1.2)
       faraday (>= 0.9, < 3, != 2.0.0)

diff --git a/helpers/application_helper.rb b/helpers/application_helper.rb
@@ -408,6 +408,9 @@ def get_ontology_and_submission
             error 404, "Ontology #{@params["acronym"]} does not have any submissions"
           end
         end
+
+        save_submission_language(submission)
+
         return ont, submission
       end
 
@@ -454,6 +457,22 @@ def naive_expiring_cache_read(key)
         return object[:object]
       end
 
+
+      def save_submission_language(submission, language_property = :naturalLanguage)
+        request_lang = RequestStore.store[:requested_lang]
+
+        return if submission.nil? || !request_lang.blank?
+
+        submission.bring(language_property) if submission.bring?(language_property)
+        collection_natural_language = submission.send(language_property) rescue nil
+        return [] if collection_natural_language.blank?
+
+        collection_natural_language = collection_natural_language.values.flatten if collection_natural_language.is_a?(Hash)
+        submissions_language = collection_natural_language.map { |natural_language| natural_language.to_s.split('/').last[0..1] }.compact.first
+
+        RequestStore.store[:requested_lang] =  submissions_language if submissions_language
+      end
+
     end
   end
 end

diff --git a/test/controllers/test_classes_controller.rb b/test/controllers/test_classes_controller.rb
@@ -8,7 +8,8 @@ def self.before_suite
                submissions_to_process: [1, 2],
                process_submission: true,
                random_submission_count: false,
-               process_options: {process_rdf: true, extract_metadata: false}
+               process_options: {process_rdf: true, extract_metadata: false},
+               file_path: "./test/data/ontology_files/BRO_v3.2.owl",
     }
     return LinkedData::SampleData::Ontology.create_ontologies_and_submissions(options)
   end
@@ -82,7 +83,6 @@ def test_notation_lookup
     assert response["@id"] == "http://bioontology.org/ontologies/BiomedicalResourceOntology.owl#Ontology_Development_and_Management"
   end
 
-
   def test_all_class_pages
     ont = Ontology.find("TEST-ONT-0").include(:acronym).first
 
@@ -99,7 +99,7 @@ def test_all_class_pages
       assert last_response.ok?
       page_response = MultiJson.load(last_response.body)
       page_response["collection"].each do |item|
-        assert_instance_of String, item["prefLabel"]
+        assert_instance_of String, item["prefLabel"], item["@id"]
         assert_instance_of String, item["@id"]
         assert_instance_of Hash, item["@context"]
         assert_instance_of Hash, item["links"]
@@ -131,6 +131,7 @@ def test_single_cls_all
     ont = Ontology.find("TEST-ONT-0").include(:acronym).first
     clss_ids = [ 'http://bioontology.org/ontologies/BiomedicalResourceOntology.owl#Molecular_Interaction',
             "http://bioontology.org/ontologies/BiomedicalResourceOntology.owl#Electron_Microscope" ]
+
     clss_ids.each do |cls_id|
       escaped_cls= CGI.escape(cls_id)
       call = "/ontologies/#{ont.acronym}/classes/#{escaped_cls}?include=all"
@@ -149,6 +150,7 @@ def test_single_cls_properties
     ont = Ontology.find("TEST-ONT-0").include(:acronym).first
     clss_ids = [ 'http://bioontology.org/ontologies/BiomedicalResourceOntology.owl#Molecular_Interaction',
             "http://bioontology.org/ontologies/BiomedicalResourceOntology.owl#Electron_Microscope" ]
+
     clss_ids.each do |cls_id|
       escaped_cls= CGI.escape(cls_id)
       call = "/ontologies/#{ont.acronym}/classes/#{escaped_cls}?include=properties"
@@ -163,7 +165,6 @@ def test_single_cls_properties
 
   def test_single_cls
     ont = Ontology.find("TEST-ONT-0").include(:acronym).first
-
     clss_ids = [ 'http://bioontology.org/ontologies/BiomedicalResourceOntology.owl#Molecular_Interaction',
             "http://bioontology.org/ontologies/BiomedicalResourceOntology.owl#Electron_Microscope" ]
 
@@ -182,13 +183,8 @@ def test_single_cls
         assert_instance_of(Array, cls["synonym"])
         assert(cls["@id"] == cls_id)
 
-        if submission_id.nil? || submission_id == 2
-          assert(cls["prefLabel"]["In version 3.2"])
-          assert(cls["definition"][0]["In version 3.2"])
-        else
-          assert(!cls["prefLabel"]["In version 3.2"])
-          assert(!cls["definition"][0]["In version 3.2"])
-        end
+        assert(cls["prefLabel"]["In version 3.2"])
+        assert(cls["definition"][0]["In version 3.2"])
 
         if cls["prefLabel"].include? "Electron"
           assert_equal(1, cls["synonym"].length)
@@ -208,7 +204,7 @@ def test_roots_for_cls
     roots = MultiJson.load(last_response.body)
     assert_includes [9, 6, 5], roots.length
     roots.each do |r|
-      assert_instance_of String, r["prefLabel"]
+      assert_instance_of String, r["prefLabel"], r["@id"]
       assert_instance_of String, r["@id"]
       assert r.include?"hasChildren"
       #By definition roots have no parents
@@ -221,7 +217,6 @@ def test_roots_for_cls
   end
 
   def test_classes_for_not_parsed_ontology
-
     ont = Ontology.find("TEST-ONT-0").include(:acronym).first
 
     #first submission was not parsed
@@ -232,7 +227,6 @@ def test_classes_for_not_parsed_ontology
   end
 
   def test_tree
-
     ont = Ontology.find("TEST-ONT-0").include(:acronym).first
 
     clss_ids = [ 'http://bioontology.org/ontologies/BiomedicalResourceOntology.owl#Molecular_Interaction',
@@ -251,7 +245,6 @@ def test_tree
   end
 
   def test_path_to_root_for_cls
-
     ont = Ontology.find("TEST-ONT-0").include(:acronym).first
     clss_ids = [ 'http://bioontology.org/ontologies/BiomedicalResourceOntology.owl#Molecular_Interaction',
             "http://bioontology.org/ontologies/BiomedicalResourceOntology.owl#Electron_Microscope" ]
@@ -264,7 +257,6 @@ def test_path_to_root_for_cls
   end
 
   def test_ancestors_for_cls
-
     ont = Ontology.find("TEST-ONT-0").include(:acronym).first
     ancestors_data = {}
     ancestors_data['http://bioontology.org/ontologies/BiomedicalResourceOntology.owl#Molecular_Interaction'] =[
@@ -456,7 +448,6 @@ def test_calls_not_found
   end
 
   def test_children_for_cls_round_trip
-
     clss_ids = [ 'http://bioontology.org/ontologies/BiomedicalResourceOntology.owl#Molecular_and_Cellular_Data',
             "http://bioontology.org/ontologies/BiomedicalResourceOntology.owl#Microscope" ]
 
@@ -513,7 +504,7 @@ def test_class_page_with_metric_count
       assert last_response.ok?
       page_response = MultiJson.load(last_response.body)
       page_response["collection"].each do |item|
-        assert_instance_of String, item["prefLabel"]
+        assert_instance_of String, item["prefLabel"], item["@id"]
         assert_instance_of String, item["@id"]
         assert_instance_of Hash, item["@context"]
         assert_instance_of Hash, item["links"]
@@ -533,4 +524,48 @@ def test_class_page_with_metric_count
     assert page_response["collection"].length == 0
   end
 
+  def test_default_multilingual
+    ont = Ontology.find("TEST-ONT-0").include(:acronym).first
+    sub = ont.latest_submission
+    sub.bring_remaining
+
+    # rdfs:label is NOT present and the prefLabel is NOT defined for the ontology or portal language
+    sub.naturalLanguage = [RDF::URI.new('http://lexvo.org/id/iso639-3/es')]
+    sub.save
+
+    get "/ontologies/#{ont.acronym}/classes/#{CGI.escape('http://bioontology.org/ontologies/Activity.owl#Community_Engagement')}"
+    assert last_response.ok?
+    page_response = MultiJson.load(last_response.body)
+    # does not contain a value in english show the generated one
+    assert_equal 'Community_Engagement', page_response["prefLabel"]
+
+    # rdfs:label is present but NOT for either the ontology or portal language
+    get "/ontologies/#{ont.acronym}/classes/#{CGI.escape('http://bioontology.org/ontologies/Activity.owl#Biospecimen_Management')}"
+    assert last_response.ok?
+    page_response = MultiJson.load(last_response.body)
+    # does not contain a value in english show the generated one
+    assert_equal 'Biospecimen_Management', page_response["prefLabel"]
+
+    # prefLabel NOT present, rdfs:label is present but NOT for either the ontology or portal language
+    get "/ontologies/#{ont.acronym}/classes/#{CGI.escape('http://bioontology.org/ontologies/Activity.owl#Gene_Therapy')}"
+    assert last_response.ok?
+    page_response = MultiJson.load(last_response.body)
+    # does not contain a value in english show the generated one
+    assert_equal 'Gene_Therapy', page_response["prefLabel"]
+
+    # prefLabel is present in the ontology language
+    sub.naturalLanguage = [RDF::URI.new('http://lexvo.org/id/iso639-3/fr')]
+    sub.save
+
+
+    get "/ontologies/#{ont.acronym}/classes/#{CGI.escape('http://bioontology.org/ontologies/Activity.owl#Biospecimen_Management')}"
+    assert last_response.ok?
+    page_response = MultiJson.load(last_response.body)
+    # show french value as specified in submission naturalLanguage
+    assert_equal 'Biospecimen Management', page_response["prefLabel"]
+
+    sub.naturalLanguage = []
+    sub.save
+  end
+
 end
diff --git a/test/data/ontology_files/BRO_v3.2.owl b/test/data/ontology_files/BRO_v3.2.owl
@@ -645,7 +645,7 @@
     <owl:Class rdf:about="&activity;Biospecimen_Management">
         <rdfs:subClassOf rdf:resource="&activity;Activity"/>
         <desc:definition rdf:datatype="&xsd;string">Activity related to the creation, use, or maintenance of a biorepository (http://en.wikipedia.org/wiki/Biorepository)</desc:definition>
-        <core:prefLabel rdf:datatype="&xsd;string">Biospecimen Management</core:prefLabel>
+        <core:prefLabel xml:lang="fr">Biospecimen Management</core:prefLabel>
     </owl:Class>
 
 
@@ -655,7 +655,6 @@
     <owl:Class rdf:about="&activity;Community_Engagement">
         <rdfs:subClassOf rdf:resource="&activity;Activity"/>
         <desc:definition rdf:datatype="&xsd;string">As defined in http://en.wikipedia.org/wiki/Community_engagement</desc:definition>
-        <core:prefLabel rdf:datatype="&xsd;string">Community Engagement</core:prefLabel>
     </owl:Class>
 
 
@@ -675,7 +674,7 @@
     <owl:Class rdf:about="&activity;Gene_Therapy">
         <rdfs:subClassOf rdf:resource="&activity;Therapeutics"/>
         <desc:definition rdf:datatype="&xsd;string">As defined in http://en.wikipedia.org/wiki/Gene_therapy</desc:definition>
-        <core:prefLabel rdf:datatype="&xsd;string">Gene Therapy</core:prefLabel>
+        <core:prefLabel xml:lang="fr">Gene Therapy</core:prefLabel>
     </owl:Class>
 
 
@@ -6328,4 +6327,3 @@ radionuclide (tracer), which is introduced into the body on a biologically activ
 
 
 <!-- Generated by the OWL API (version 3.4.2) http://owlapi.sourceforge.net -->
-