Fix: issue when generating multilingual missing labels (#164)

* fix the issue when generating multilingual missing labels * fixed the issues revealed by the failing unit tests * handle the case that in ontoportal-lirmm branch we use "@none" not "none"
ontoportal-lirmm · Oct 24, 2024 · 793c4fd · 793c4fd
1 parent 0a45655
commit 793c4fd
Show file tree

Hide file tree

Showing 7 changed files with 26,209 additions and 42 deletions.
diff --git a/Gemfile.lock b/Gemfile.lock
@@ -1,6 +1,6 @@
 GIT
   remote: https://github.com/ontoportal-lirmm/goo.git
-  revision: a95245b8c964431505ca6315907440996c59a00d
+  revision: f8ac7b00e8d8b46d1eea04de014175525c1cdd83
   branch: development
   specs:
     goo (0.0.2)
@@ -60,7 +60,7 @@ GEM
       launchy (>= 2.1, < 4.0)
       mail (~> 2.7)
     eventmachine (1.2.7)
-    faraday (1.10.3)
+    faraday (1.10.4)
       faraday-em_http (~> 1.0)
       faraday-em_synchrony (~> 1.0)
       faraday-excon (~> 1.1)
@@ -96,7 +96,7 @@ GEM
     json-ld (3.0.2)
       multi_json (~> 1.12)
       rdf (>= 2.2.8, < 4.0)
-    jwt (2.9.0)
+    jwt (2.9.3)
       base64
     language_server-protocol (3.17.0.3)
     launchy (3.0.1)
@@ -113,9 +113,10 @@ GEM
       net-pop
       net-smtp
     method_source (1.1.0)
-    mime-types (3.5.2)
+    mime-types (3.6.0)
+      logger
       mime-types-data (~> 3.2015)
-    mime-types-data (3.2024.0903)
+    mime-types-data (3.2024.1001)
     mini_mime (1.1.5)
     minitest (4.7.5)
     minitest-reporters (0.14.24)
@@ -127,7 +128,7 @@ GEM
     multipart-post (2.4.1)
     net-http-persistent (4.0.4)
       connection_pool (~> 2.2)
-    net-imap (0.4.16)
+    net-imap (0.4.17)
       date
       net-protocol
     net-pop (0.1.2)
@@ -188,10 +189,10 @@ GEM
       http-cookie (>= 1.0.2, < 2.0)
       mime-types (>= 1.16, < 4.0)
       netrc (~> 0.8)
-    rexml (3.3.7)
+    rexml (3.3.8)
     rsolr (1.1.2)
       builder (>= 2.1.2)
-    rubocop (1.66.1)
+    rubocop (1.67.0)
       json (~> 2.3)
       language_server-protocol (>= 3.17.0)
       parallel (~> 1.10)
@@ -228,7 +229,7 @@ GEM
     unicode-display_width (2.6.0)
     uuid (2.3.9)
       macaddr (~> 1.0)
-    webmock (3.23.1)
+    webmock (3.24.0)
       addressable (>= 2.8.0)
       crack (>= 0.3.2)
       hashdiff (>= 0.4.0, < 2.0.0)

diff --git a/...ntologies_linked_data/services/submission_process/operations/submission_missing_labels.rb b/...ntologies_linked_data/services/submission_process/operations/submission_missing_labels.rb
@@ -10,6 +10,7 @@ def process(logger, options = {})
 
       def handle_missing_labels(file_path, logger)
         callbacks = {
+          include_languages: true,
           missing_labels: {
             op_name: 'Missing Labels Generation',
             required: true,
@@ -59,6 +60,11 @@ def loop_classes(logger, raw_paging, submission, callbacks)
         size = 2500
         count_classes = 0
         acr = submission.id.to_s.split("/")[-1]
+
+        # include all languages in attributes of classes if asked for
+        incl_lang = callbacks.delete(:include_languages)
+        RequestStore.store[:requested_lang] = :ALL if incl_lang
+
         operations = callbacks.values.map { |v| v[:op_name] }.join(", ")
 
         time = Benchmark.realtime do
@@ -78,9 +84,7 @@ def loop_classes(logger, raw_paging, submission, callbacks)
           # 1. init artifacts hash if not explicitly passed in the callback
           # 2. determine if class-level iteration is required
           callbacks.each { |_, callback| callback[:artifacts] ||= {};
-          if callback[:caller_on_each]
-            iterate_classes = true
-          end }
+          iterate_classes = true if callback[:caller_on_each] }
 
           process_callbacks(logger, callbacks, :caller_on_pre) {
             |callable, callback| callable.call(callback[:artifacts], logger, paging) }
@@ -95,7 +99,7 @@ def loop_classes(logger, raw_paging, submission, callbacks)
             page_len = page_classes.length
 
             # nothing retrieved even though we're expecting more records
-            if total_pages > 0 && page_classes.empty? && (prev_page_len == -1 || prev_page_len == size)
+            if total_pages.positive? && page_classes.empty? && (prev_page_len == -1 || prev_page_len == size)
               j = 0
               num_calls = LinkedData.settings.num_retries_4store
 
@@ -117,7 +121,7 @@ def loop_classes(logger, raw_paging, submission, callbacks)
             end
 
             if page_classes.empty?
-              if total_pages > 0
+              if total_pages.positive?
                 logger.info("The number of pages reported for #{acr} - #{total_pages} is higher than expected #{page - 1}. Completing #{operations}...")
               else
                 logger.info("Ontology #{acr} contains #{total_pages} pages...")
@@ -160,6 +164,7 @@ def loop_classes(logger, raw_paging, submission, callbacks)
             @submission.save
           end
         end
+        RequestStore.store[:requested_lang] = nil if incl_lang
       end
 
       def generate_missing_labels_pre(artifacts = {}, logger, paging)
@@ -181,49 +186,68 @@ def generate_missing_labels_pre_page(artifacts = {}, logger, paging, page_classe
       end
 
       def generate_missing_labels_each(artifacts = {}, logger, paging, page_classes, page, c)
-        prefLabel = nil
-
-        if c.prefLabel.nil?
-          rdfs_labels = c.label
+        pref_label = nil
+        portal_lang = Goo.portal_language
+        pref_label_lang = c.prefLabel(include_languages: true)
+        no_default_pref_label = pref_label_lang.nil? || (pref_label_lang.keys & [portal_lang, :none, '@none']).empty?
 
-          if rdfs_labels && rdfs_labels.length > 1 && c.synonym.length > 0
-            rdfs_labels = (Set.new(c.label) - Set.new(c.synonym)).to_a.first
+        if no_default_pref_label
+          lang_rdfs_labels = c.label(include_languages: true)
 
-            rdfs_labels = c.label if rdfs_labels.nil? || rdfs_labels.length == 0
+          # Set lang_rdfs_labels to { none: [] } if empty or no match for default label
+          if Array(lang_rdfs_labels).empty? || (lang_rdfs_labels.keys & [portal_lang, :none, '@none']).empty?
+            lang_rdfs_labels = { none: [] }
           end
 
-          rdfs_labels = [rdfs_labels] if rdfs_labels and not (rdfs_labels.instance_of? Array)
-          label = nil
+          lang_rdfs_labels.each do |lang, rdfs_labels|
+            # Remove synonyms from rdfs_labels if there are multiple labels and synonyms exist
+            if rdfs_labels&.length.to_i > 1 && c.synonym.present?
+              rdfs_labels = (Set.new(c.label) - Set.new(c.synonym)).to_a.first || c.label
+            end
 
-          if rdfs_labels && rdfs_labels.length > 0
-            label = rdfs_labels[0]
-          else
-            label = LinkedData::Utils::Triples.last_iri_fragment c.id.to_s
+            # Ensure rdfs_labels is an array
+            rdfs_labels = Array(rdfs_labels) if rdfs_labels && !rdfs_labels.is_a?(Array)
+
+            # Select the label: either the minimal sorted label or the last fragment of the IRI
+            label = rdfs_labels&.min || LinkedData::Utils::Triples.last_iri_fragment(c.id.to_s)
+
+            # Set language to nil for :none and assign pref_label
+            lang = nil if lang.eql?(:none) || lang.to_s.eql?('@none')
+            pref_label = label if lang.nil? || lang.eql?(portal_lang)
+            pref_label ||= label
+
+            artifacts[:label_triples] << LinkedData::Utils::Triples.label_for_class_triple(
+              c.id, Goo.vocabulary(:metadata_def)[:prefLabel], pref_label, lang
+            )
           end
-          artifacts[:label_triples] << LinkedData::Utils::Triples.label_for_class_triple(
-            c.id, Goo.vocabulary(:metadata_def)[:prefLabel], label)
-          prefLabel = label
+        elsif pref_label_lang
+          pref_label = c.prefLabel
         else
-          prefLabel = c.prefLabel
+          pref_label = LinkedData::Utils::Triples.last_iri_fragment(c.id.to_s)
         end
 
-        if @submission.ontology.viewOf.nil?
-          loomLabel = LinkedData::Models::OntologySubmission.loom_transform_literal(prefLabel.to_s)
+        # Handle loom transformation if ontology is not a view
+        unless @submission.ontology.viewOf
+          loom_label = LinkedData::Models::OntologySubmission.loom_transform_literal(pref_label.to_s)
 
-          if loomLabel.length > 2
+          if loom_label.length > 2
             artifacts[:mapping_triples] << LinkedData::Utils::Triples.loom_mapping_triple(
-              c.id, Goo.vocabulary(:metadata_def)[:mappingLoom], loomLabel)
+              c.id, Goo.vocabulary(:metadata_def)[:mappingLoom], loom_label
+            )
           end
+
           artifacts[:mapping_triples] << LinkedData::Utils::Triples.uri_mapping_triple(
-            c.id, Goo.vocabulary(:metadata_def)[:mappingSameURI], c.id)
+            c.id, Goo.vocabulary(:metadata_def)[:mappingSameURI], c.id
+          )
         end
+
       end
 
       def generate_missing_labels_post_page(artifacts = {}, logger, paging, page_classes, page)
         rest_mappings = LinkedData::Mappings.migrate_rest_mappings(@submission.ontology.acronym)
         artifacts[:mapping_triples].concat(rest_mappings)
 
-        if artifacts[:label_triples].length > 0
+        if artifacts[:label_triples].length.positive?
           logger.info("Asserting #{artifacts[:label_triples].length} labels in " +
                         "#{@submission.id.to_ntriples}")
           logger.flush
@@ -239,7 +263,7 @@ def generate_missing_labels_post_page(artifacts = {}, logger, paging, page_class
           logger.flush
         end
 
-        if artifacts[:mapping_triples].length > 0
+        if artifacts[:mapping_triples].length.positive?
           logger.info("Asserting #{artifacts[:mapping_triples].length} mappings in " +
                         "#{@submission.id.to_ntriples}")
           logger.flush
@@ -266,4 +290,3 @@ def generate_missing_labels_post(artifacts = {}, logger, pagging)
 
   end
 end
-
diff --git a/lib/ontologies_linked_data/utils/triples.rb b/lib/ontologies_linked_data/utils/triples.rb
@@ -69,10 +69,17 @@ def self.rdf_for_custom_properties(ont_sub)
         return (triples.join "\n")
       end
 
-      def self.label_for_class_triple(class_id,property,label)
+      def self.label_for_class_triple(class_id, property, label, language=nil)
         label = label.to_s.gsub('\\','\\\\\\\\')
         label = label.gsub('"','\"')
-        return triple(class_id,property,RDF::Literal.new(label, :datatype => RDF::XSD.string))
+        params = { datatype: RDF::XSD.string }
+        lang = language.to_s.downcase
+
+        if !lang.empty? && lang.to_sym != :none && !lang.to_s.eql?('@none')
+          params[:datatype] = RDF.langString
+          params[:language] = lang.to_sym
+        end
+        triple(class_id, property, RDF::Literal.new(label, params))
       end
 
       def self.generated_label(class_id, existing_label)