Feature: Add canonical portal for duplicate ontologies

Gemfile

@@ -100,8 +100,8 @@ gem "flag-icons-rails", "~> 3.4"
 
 # Custom API client
 gem 'ontologies_api_client', git: 'https://github.com/ontoportal-lirmm/ontologies_api_ruby_client.git', branch: 'development'
-
 # Ruby 2.7.8 pinned gems (to remove when migrating to Ruby >= 3.0)
+
 gem 'ffi', '~> 1.16.3'
 gem 'net-ftp', '~> 0.2.0', require: false
 gem 'net-http', '~> 0.3.2'

app/controllers/concerns/search_aggregator.rb

@@ -31,12 +31,16 @@ def aggregate_results(query, results)
 
     all_ontologies = LinkedData::Client::Models::Ontology.all(include: 'acronym,name', include_views: true, display_links: false, display_context: false)
 
+    all_submissions = LinkedData::Client::Models::OntologySubmission.all(include: 'pullLocation', include_views: true, display_links: false, display_context: false)
+
     search_results = grouped_results.map do |group|
       format_search_result(group, all_ontologies)
     end
 
     search_results = merge_sort_federated_results(query, search_results) if federation_enabled?
 
+    search_results = apply_canonical_portal(search_results, all_submissions)
+
     search_results
   end
 
@@ -91,7 +95,6 @@ def search_result_elem(class_object, ontology_acronym, title)
     result
   end
 
-
   def ontology_name_acronym(ontologies, selected_acronym)
     ontology = ontologies.select { |x| x.acronym.eql?(selected_acronym.split('/').last) }.first
     "#{ontology.name} (#{ontology.acronym})" if ontology

app/controllers/concerns/submission_filter.rb

@@ -71,7 +71,7 @@ def merge_by_acronym(submissions)
       if ontologies.size.eql?(1)
         ontology = ontologies.first
       else
-        ontology = ontologies.select { |x| helpers.internal_ontology?(x[:id]) }.first || ontologies.first
+        ontology = canonical_portal_ontology(ontologies)
       end
 
       ontology[:sources] = ontologies.map { |x| x[:id] }
@@ -80,6 +80,10 @@ def merge_by_acronym(submissions)
     merged_submissions
   end
 
+  def canonical_portal_ontology(ontologies)
+    ontologies.select { |x| helpers.internal_ontology?(x[:id]) }.first || ontology_from_portal(ontologies, canonical_portal(ontologies))
+  end
+
   def filter_submissions(ontologies, query:, status:, show_views:, private_only:, languages:, page_size:, formality_level:, is_of_type:, groups:, categories:, formats:)
     submissions = LinkedData::Client::Models::OntologySubmission.all(include: BROWSE_ATTRIBUTES.join(','), also_include_views: true, display_links: false, display_context: false)
 

app/helpers/federation_helper.rb

@@ -124,4 +124,55 @@ def federation_external_class?(class_object)
   end
 
 
+  def canonical_portal(ontologies)
+    portal_counts = Hash.new(0)
+    # Count occurrences of each portal in the pull_location URL
+    ontologies.each do |ontology|
+      federated_portals.keys.each do |portal|
+        if ontology[:pullLocation]&.include?(portal.to_s)
+          portal_counts[portal] += 1
+        end
+      end
+    end
+    # Determine the portal with the most occurrences
+    canonical_portal = portal_counts.max_by { |_, count| count }&.first
+
+    canonical_portal
+  end
+
+  def apply_canonical_portal(search_results, all_submissions)
+    search_results.each do |result|
+      next if result[:root][:portal_name].nil? || result[:root][:other_portals].blank?
+
+      root_link = result[:root][:link].split('?').first
+      candidates = [result[:root][:link]] + result[:root][:other_portals].map { |p| p[:link].split('?').first }
+
+      portal_counts = Hash.new(0)
+      candidates.each do |candidate|
+        submission = all_submissions.find { |s| s.id.split('/')[0...-2].join('/') == candidate }
+
+        federated_portals.keys.each do |portal|
+          portal_counts[portal] += 1 if submission&.include?(portal.to_s)
+        end
+      end
+
+      canonical_portal = portal_counts.max_by { |_, count| count }&.first
+      next if canonical_portal.nil? || result[:root][:portal_name].eql?(canonical_portal.to_s)
+
+      canonical_portal_result = result[:root][:other_portals].find { |r| r[:name] == canonical_portal.to_s }
+      swap_portal_attributes(result[:root], canonical_portal_result) if canonical_portal_result
+    end
+  end
+
+  def swap_portal_attributes(root_portal, new_portal)
+    root_portal[:link], new_portal[:link] = new_portal[:link], root_portal[:link]
+    root_portal[:portal_name], new_portal[:portal_name] = new_portal[:portal_name], root_portal[:portal_name]
+    root_portal[:portal_color], new_portal[:portal_color] = new_portal[:portal_color], root_portal[:portal_color]
+    root_portal[:portal_light_color], new_portal[:portal_light_color] = new_portal[:portal_light_color], root_portal[:portal_light_color]
+  end
+
+  def ontology_from_portal(ontologies, portal)
+    ontologies.select{|o| o[:id].include?(portal.to_s)}.first
+  end
+
 end