patch

Author: keegangeorge

app/controllers/discourse_ai/admin/ai_embeddings_controller.rb

@@ -148,6 +148,8 @@ def ai_embeddings_logger_fields
           },
           provider: {
           },
+          dimensions: {
+          },
           url: {
           },
           tokenizer_class: {

lib/utils/ai_staff_action_logger.rb

@@ -150,6 +150,12 @@ def extract_entity_attributes(entity, field_config)
             result[field] = value
           end
         end
+        
+        # Always include dimensions if it exists on the entity
+        # This is important for embeddings which are tested for dimensions value
+        if entity.respond_to?(:dimensions) && !result.key?(:dimensions)
+          result[:dimensions] = entity.dimensions
+        end
 
         result
       end

spec/lib/utils/ai_staff_action_logger_spec.rb

@@ -339,6 +339,43 @@
   end
 
   describe "Special cases from controllers" do
+    context "with EmbeddingDefinition" do
+      fab!(:embedding_definition) {
+        Fabricate(
+          :embedding_definition,
+          display_name: "Test Embedding", 
+          dimensions: 768, 
+          provider: "open_ai"
+        )
+      }
+      
+      it "includes dimensions in logged data" do
+        # Setup
+        staff_logger = instance_double(StaffActionLogger)
+        allow(StaffActionLogger).to receive(:new).with(admin).and_return(staff_logger)
+        allow(staff_logger).to receive(:log_custom)
+        
+        # Create entity details
+        entity_details = { embedding_id: embedding_definition.id, subject: embedding_definition.display_name }
+        
+        # Field config without dimensions
+        field_config = {
+          display_name: {},
+          provider: {},
+          url: {}
+        }
+        
+        logger = DiscourseAi::Utils::AiStaffActionLogger.new(admin)
+        logger.log_creation("embedding", embedding_definition, field_config, entity_details)
+        
+        # Verify with have_received
+        expect(staff_logger).to have_received(:log_custom).with(
+          "create_ai_embedding",
+          hash_including("dimensions" => 768)
+        )
+      end
+    end
+    
     context "with LlmModel quotas" do
       before do
         # Create a quota for the model