Store entities in Sets for fast lookups. #27

Author: fbacall

lib/ro_crate/model/crate.rb

@@ -1,3 +1,5 @@
+require 'set'
+
 module ROCrate
   ##
   # A Ruby abstraction of an RO-Crate.
@@ -20,8 +22,8 @@ def self.format_local_id(id)
     ##
     # Initialize an empty RO-Crate.
     def initialize(id = IDENTIFIER, properties = {})
-      @data_entities = []
-      @contextual_entities = []
+      @data_entities = Set.new
+      @contextual_entities = Set.new
       super(self, nil, id, properties)
     end
 
@@ -144,8 +146,8 @@ def add_organization(id, properties = {})
     # @return [Entity] the entity itself, or a clone of the entity "owned" by this crate.
     def add_contextual_entity(entity)
       entity = claim(entity)
-      contextual_entities.delete(entity) # Remove (then re-add) the entity if it exists
-      contextual_entities.push(entity)
+      contextual_entities.delete?(entity) # Remove (then re-add) the entity if it exists
+      contextual_entities.add(entity)
       entity
     end
 
@@ -156,8 +158,8 @@ def add_contextual_entity(entity)
     # @return [Entity] the entity itself, or a clone of the entity "owned" by this crate.
     def add_data_entity(entity)
       entity = claim(entity)
-      data_entities.delete(entity) # Remove (then re-add) the entity if it exists
-      data_entities.push(entity)
+      data_entities.delete?(entity) # Remove (then re-add) the entity if it exists
+      data_entities.add(entity)
       entity
     end
 
@@ -191,7 +193,7 @@ def preview=(preview)
     #
     # @return [Array<Entity>]
     def entities
-      default_entities | data_entities | contextual_entities
+      default_entities | data_entities.to_a | contextual_entities.to_a
     end
 
     ##
@@ -249,7 +251,7 @@ def payload
       # file data entities. This ensures in the case of a conflict, the more "specific" data entities take priority.
       entries = own_payload
       non_self_entities = default_entities.reject { |e| e == self }
-      sorted_entities = (non_self_entities | data_entities).sort_by { |e| e.is_a?(ROCrate::Directory) ? 0 : 1 }
+      sorted_entities = (non_self_entities | data_entities.to_a).sort_by { |e| e.is_a?(ROCrate::Directory) ? 0 : 1 }
 
       sorted_entities.each do |entity|
         entity.payload.each do |path, entry|
@@ -277,7 +279,7 @@ def delete(entity, remove_orphaned: true)
       entity = dereference(entity) if entity.is_a?(String)
       return unless entity
 
-      deleted = data_entities.delete(entity) || contextual_entities.delete(entity)
+      deleted = data_entities.delete?(entity) || contextual_entities.delete?(entity)
 
       if deleted && remove_orphaned
         crate_entities = crate.linked_entities(deep: true)

lib/ro_crate/model/entity.rb

@@ -251,7 +251,7 @@ def linked_entities(deep: false, linked: {})
         value = properties[key] # We're doing this to call the JSONLDHash#[] method which wraps
         value = [value] if value.is_a?(JSONLDHash)
 
-        if value.is_a?(Array)
+        if value.respond_to?(:each)
           value.each do |v|
             if v.is_a?(JSONLDHash) && !linked.key?(v['@id'])
               entity = v.dereference

test/crate_test.rb

@@ -171,7 +171,8 @@ class CrateTest < Test::Unit::TestCase
     crate.author = bob
     info.author = bob
 
-    assert_equal [bob], crate.contextual_entities
+    assert_includes crate.contextual_entities, bob
+    assert_equal 1, crate.contextual_entities.length
     assert_equal bob, info.author
     assert_equal bob, crate.author
   end