Feature: Add canonical portal for duplicate ontologies (#770)

* method to get the canonical portal for an ontology using pullLocation

* display federated ontologies in browse page based on the canonical portal

* clean canonical portal function in federation helper

* extract canonical portal ontology choice in a function for clarity

* display search page results based on the canonical portal of the ontology

* move federation canonical to federation helper

* clean swap portal attributes function in federation helper

* fix and clean apply canonical portal function

* extract count portals into a separate function from external_canonical_ontology_portal function + apply_canonical_portal function

* put again search canonical logic the search_aggregator file and refactor the code

* do the search canonical logic only if enabled

* fix rest_hostname raising an exception that REST_URI does not exist

---------

Co-authored-by: Syphax bouazzouni <[email protected]>

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

@@ -31,11 +31,15 @@ def aggregate_results(query, results)
 
     all_ontologies = LinkedData::Client::Models::Ontology.all(include: 'acronym,name', 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?
+    if federation_enabled?
+      search_results = merge_sort_federated_results(query, search_results) if federation_enabled?
+      search_results = swap_canonical_portal_results_first(search_results)
+    end
 
     search_results
   end
@@ -251,7 +255,13 @@ def merge_federated_results(search_results)
         if (element[:root][:ontology_acronym] == reuse[:root][:ontology_acronym]) && (element[:root][:uri] == reuse[:root][:uri])
           portal_name = reuse[:root][:portal_name]
           link = reuse[:root][:link]
-          element[:root][:other_portals] << {name: portal_name, color: federated_portal_color(portal_name), light_color: federated_portal_light_color(portal_name), link: link}
+          element[:root][:other_portals] << {
+            name: portal_name,
+            color: federated_portal_color(portal_name),
+            light_color: federated_portal_light_color(portal_name),
+            link: link,
+            ontology_id: reuse[:root][:ontology_id]
+          }
           true
         else
           false
@@ -260,6 +270,36 @@ def merge_federated_results(search_results)
     end
   end
 
+  def swap_canonical_portal_results_first(search_results)
+    all_submissions = LinkedData::Client::Models::OntologySubmission.all(include: 'pullLocation', include_views: true, display_links: false, display_context: false)
+
+    search_results.each do |result|
+      next if result[:root][:portal_name].nil? || result[:root][:other_portals].blank?
+
+      result_ontology_ids = [result[:root][:ontology_id]] + result[:root][:other_portals].map { |p| p[:ontology_id] }
+
+      result_submissions = all_submissions.select do |submission|
+        result_ontology_ids.any? { |ontology_id| submission.id.include?(ontology_id) }
+      end
+
+      canonical_portal = most_referred_portal(result_submissions)
+      is_internal_ontology = result[:root][:portal_name].eql?(canonical_portal.to_s)
+
+      next if canonical_portal.nil? || is_internal_ontology
+
+      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
+    search_results
+  end
+
+
+  def swap_portal_attributes(root_portal, new_portal)
+    [:link, :portal_name, :portal_color, :portal_light_color].each do |attribute|
+      root_portal[attribute], new_portal[attribute] = new_portal[attribute], root_portal[attribute]
+    end
+  end
+
   def sort_results_by_string_similarity(query, search_results)
     search_results = search_results.sort_by do |entry|
       root_similarity = String::Similarity.cosine(query.downcase, entry[:root][:title].split('-').first.gsub(" ", "").downcase)

app/controllers/concerns/submission_filter.rb

@@ -70,18 +70,14 @@ def ontologies_with_filters_url(filters, page: 1, count: false)
   def merge_by_acronym(submissions)
     merged_submissions = []
     submissions.group_by { |x| x[:ontology]&.acronym }.each do |acronym, ontologies|
-      if ontologies.size.eql?(1)
-        ontology = ontologies.first
-      else
-        ontology = ontologies.select { |x| helpers.internal_ontology?(x[:id]) }.first || ontologies.first
-      end
-
+      ontology = canonical_ontology(ontologies)
       ontology[:sources] = ontologies.map { |x| x[:id] }
       merged_submissions << ontology
     end
     merged_submissions
   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

@@ -135,6 +135,18 @@ def federation_external_class?(class_object)
     !class_object.links['self'].include?($REST_URL)
   end
 
+  def canonical_ontology(ontologies)
+    if ontologies.size.eql?(1)
+      ontologies.first
+    else
+      internal_ontology = ontologies.select { |x| helpers.internal_ontology?(x[:id]) }.first
+      if internal_ontology
+        internal_ontology
+      else
+        external_canonical_ontology_portal(ontologies)
+      end
+    end
+  end
 
   def federation_portal_status(portal_name: nil)
     Rails.cache.fetch("federation_portal_up_#{portal_name}", expires_in: 2.hours) do
@@ -191,6 +203,7 @@ def init_federation_portals_status
       federation_input_chips
     end
   end
+
   def federated_search_counts(search_results)
     ids = search_results.map do |result|
       result.dig(:root, :ontology_id) || rest_url
@@ -219,4 +232,20 @@ def counts_ontology_ids_by_portal_name(portals_ids)
 
     counts
   end
+
+  def external_canonical_ontology_portal(ontologies)
+    canonical_portal = most_referred_portal(ontologies)
+    ontologies.select{|o| o[:id].include?(canonical_portal.to_s)}.first
+  end
+
+  def most_referred_portal(ontology_submissions)
+    portal_counts = Hash.new(0)
+    ontology_submissions.each do |submission|
+      federated_portals.keys.each do |portal|
+        portal_counts[portal] += 1 if submission[:pullLocation]&.include?(portal.to_s)
+      end
+    end
+    portal_counts.max_by { |_, count| count }&.first
+  end
+
 end

app/helpers/urls_helper.rb

@@ -5,7 +5,7 @@ def url_to_endpoint(url)
         endpoint
     end
     def rest_hostname
-        extract_hostname(REST_URI)
+        extract_hostname($REST_URL)
     end
 
     def extract_hostname(url)