Make page batch jobs nested classes of their counterparts

Author: rossta

app/jobs/pages/analyze_topics_job.rb

@@ -3,6 +3,16 @@
 
 module Pages
   class AnalyzeTopicsJob < ApplicationJob
+    class Batch < ApplicationJob
+      queue_as :default
+
+      def perform
+        Page.published.where.missing(:topics).find_each do |page|
+          Pages::AnalyzeTopicsJob.perform_later(page)
+        end
+      end
+    end
+
     queue_as :default
 
     def perform(page)

app/jobs/pages/batch_analyze_topics_job.rb

@@ -5,9 +5,9 @@ class BatchUpsertPagesJob < ApplicationJob
     def perform(limit: nil)
       Page.upsert_collection_from_sitepress!(limit: limit)
 
-      Pages::RefreshSearchIndexJob.perform_later
-      Pages::BatchAnalyzeTopicsJob.perform_later
-      Pages::BatchEmbeddingJob.perform_later
+      Pages::RefreshSearchIndexJob::Batch.perform_later
+      Pages::AnalyzeTopicsJob::Batch.perform_later
+      Pages::EmbeddingJob::Batch.perform_later
     end
   end
 end

app/jobs/pages/batch_embedding_job.rb

@@ -1,5 +1,15 @@
 module Pages
   class EmbeddingJob < ApplicationJob
+    class Batch < ApplicationJob
+      queue_as :default
+
+      def perform
+        Page.published.where.missing(:page_embedding).find_each do |page|
+          Pages::EmbeddingJob.perform_later(page)
+        end
+      end
+    end
+
     queue_as :default
 
     def perform(page)

app/jobs/pages/batch_upsert_pages_job.rb

@@ -1,7 +1,19 @@
 module Pages
   class RefreshSearchIndexJob < ApplicationJob
-    def perform
-      Page.published.find_each(&:update_in_search_index)
+    class Batch < ApplicationJob
+      queue_as :default
+
+      def perform
+        Page.published.find_each do |page|
+          Pages::RefreshSearchIndexJob.perform_later(page)
+        end
+      end
+    end
+
+    queue_as :default
+
+    def perform(page)
+      page.update_in_search_index
     end
   end
 end

app/jobs/pages/embedding_job.rb

@@ -28,4 +28,30 @@
       expect(page.reload.topics.pluck(:name)).to include("Ruby on Rails")
     end
   end
+
+  describe Pages::AnalyzeTopicsJob::Batch do
+    it "doesn’t blow up" do
+      allow(SitepressPage).to receive(:render_html).and_return(Faker::HTML.random)
+
+      topics = FactoryBot.create_list(:topic, 2)
+
+      page_1, page_2, page_3 = Page.upsert_collection_from_sitepress!(limit: 3).each do |page|
+        page.touch(:published_at)
+      end
+
+      # articles without topics
+      page_1.update!(topics: [])
+      page_2.update!(topics: [])
+
+      # articles with topics
+      page_3.update!(topics: topics)
+
+      # not articles
+      FactoryBot.create_list(:page, 3)
+
+      expect(Pages::AnalyzeTopicsJob).to receive(:perform_later).exactly(2).times
+
+      described_class.perform_now
+    end
+  end
 end

app/jobs/pages/refresh_search_index_job.rb

@@ -28,12 +28,14 @@
   end
 
   it "enqueues refresh search index job and analyze topics job" do
-    allow(Pages::RefreshSearchIndexJob).to receive(:perform_later)
-    allow(Pages::BatchAnalyzeTopicsJob).to receive(:perform_later)
+    allow(Pages::RefreshSearchIndexJob::Batch).to receive(:perform_later)
+    allow(Pages::AnalyzeTopicsJob::Batch).to receive(:perform_later)
+    allow(Pages::EmbeddingJob::Batch).to receive(:perform_later)
 
     described_class.perform_now
 
-    expect(Pages::RefreshSearchIndexJob).to have_received(:perform_later)
-    expect(Pages::BatchAnalyzeTopicsJob).to have_received(:perform_later)
+    expect(Pages::EmbeddingJob::Batch).to have_received(:perform_later)
+    expect(Pages::RefreshSearchIndexJob::Batch).to have_received(:perform_later)
+    expect(Pages::AnalyzeTopicsJob::Batch).to have_received(:perform_later)
   end
 end