Make page batch jobs nested classes of their counterparts

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_upsert_pages_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/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/refresh_search_index_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

spec/jobs/pages/analyze_topics_job_spec.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

spec/jobs/pages/batch_upsert_pages_job_spec.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

spec/jobs/pages/embedding_job_spec.rb

@@ -32,4 +32,22 @@
     expect(page_embedding.embedding).to be_a(Array)
     expect(page_embedding.embedding.length).to eq(PageEmbedding::OPENAI_EMBEDDING_LENGTH)
   end
+
+  describe Pages::EmbeddingJob::Batch do
+    it "doesn’t blow up" do
+      FactoryBot.create(:page, :published)
+      expect(Pages::EmbeddingJob).to receive(:perform_later).once
+
+      described_class.perform_now
+    end
+
+    it "doesn’t blow up when page has embedding" do
+      page = FactoryBot.create(:page, :published)
+      FactoryBot.create(:page_embedding, id: page.id)
+
+      expect(Pages::EmbeddingJob).not_to receive(:perform_later)
+
+      described_class.perform_now
+    end
+  end
 end

spec/jobs/pages/refresh_search_index_job_spec.rb

@@ -1,20 +1,24 @@
 require "rails_helper"
 
 RSpec.describe Pages::RefreshSearchIndexJob, type: :job do
-  it "doesn’t blow up when indexing the current pages" do
-    Page.upsert_collection_from_sitepress!
+  describe Pages::RefreshSearchIndexJob::Batch do
+    it "doesn’t blow up when indexing the current pages" do
+      Page.upsert_collection_from_sitepress!
+      perform_enqueued_jobs
 
-    expect { described_class.perform_now }.not_to raise_error
-  end
+      expect { Pages::RefreshSearchIndexJob::Batch.perform_now }.not_to raise_error
+    end
 
-  it "indexes the right stuff" do
-    page_1 = FactoryBot.create(:page, :published, request_path: "/articles/introducing-joy-of-rails")
-    page_2 = FactoryBot.create(:page, :published, request_path: "/articles/custom-color-schemes-with-ruby-on-rails")
-    page_3 = FactoryBot.create(:page, :unpublished, request_path: "/articles/joy-of-rails-2")
+    it "indexes the right stuff" do
+      page_1 = FactoryBot.create(:page, :published, request_path: "/articles/introducing-joy-of-rails")
+      page_2 = FactoryBot.create(:page, :published, request_path: "/articles/custom-color-schemes-with-ruby-on-rails")
+      page_3 = FactoryBot.create(:page, :unpublished, request_path: "/articles/joy-of-rails-2")
 
-    described_class.perform_now
+      Pages::RefreshSearchIndexJob::Batch.perform_now
+      perform_enqueued_jobs
 
-    expect(Page.join_search_index).to include(page_1, page_2)
-    expect(Page.join_search_index).not_to include(page_3)
+      expect(Page.join_search_index).to include(page_1, page_2)
+      expect(Page.join_search_index).not_to include(page_3)
+    end
   end
 end

spec/models/page/searchable_spec.rb

@@ -4,14 +4,14 @@
   describe ".search" do
     it "allows searching for pages" do
       page = FactoryBot.create(:page, :published, request_path: "/")
-      Pages::RefreshSearchIndexJob.perform_now
+      page.update_in_search_index
 
       expect(Page.search("Joy of Rails")).to include page
     end
 
     it "works after rebuilding index" do
       page = FactoryBot.create(:page, :published, request_path: "/")
-      Pages::RefreshSearchIndexJob.perform_now
+      page.update_in_search_index
 
       Page.find_each(&:update_in_search_index)
 
@@ -23,7 +23,8 @@
     it "orders search results by rank" do
       page_1 = FactoryBot.create(:page, :published, request_path: "/articles/custom-color-schemes-with-ruby-on-rails")
       page_2 = FactoryBot.create(:page, :published, request_path: "/articles/mastering-custom-configuration-in-rails")
-      Pages::RefreshSearchIndexJob.perform_now
+      page_1.update_in_search_index
+      page_2.update_in_search_index
 
       search = Page.search("custom con*")
 
@@ -33,8 +34,8 @@
 
   describe ".with_snippets" do
     it "orders search results by rank" do
-      FactoryBot.create(:page, :published, request_path: "/articles/custom-color-schemes-with-ruby-on-rails")
-      Pages::RefreshSearchIndexJob.perform_now
+      page = FactoryBot.create(:page, :published, request_path: "/articles/custom-color-schemes-with-ruby-on-rails")
+      page.update_in_search_index
 
       page = Page.search("Color*").with_snippets.first
 

spec/requests/search_spec.rb

@@ -35,8 +35,8 @@
     end
 
     it "renders the search results without query" do
-      FactoryBot.create(:page, :published, request_path: "/pwa-showcase")
-      Pages::RefreshSearchIndexJob.perform_now
+      article = FactoryBot.create(:page, :published, request_path: "/pwa-showcase")
+      article.update_in_search_index
 
       get search_path
 
@@ -46,9 +46,10 @@
     end
 
     it "renders the search results with query as turbo stream" do
-      FactoryBot.create(:page, :published, request_path: "/pwa-showcase")
-      FactoryBot.create(:page, :published, request_path: "/articles/introducing-joy-of-rails")
-      Pages::RefreshSearchIndexJob.perform_now
+      article1 = FactoryBot.create(:page, :published, request_path: "/pwa-showcase")
+      article2 = FactoryBot.create(:page, :published, request_path: "/articles/introducing-joy-of-rails")
+      article1.update_in_search_index
+      article2.update_in_search_index
 
       get search_path(format: :turbo_stream), params: {query: "Progressive Web Apps"}
 
@@ -63,9 +64,10 @@
     end
 
     it "renders the search results with query" do
-      FactoryBot.create(:page, :published, request_path: "/pwa-showcase")
-      FactoryBot.create(:page, :published, request_path: "/articles/introducing-joy-of-rails")
-      Pages::RefreshSearchIndexJob.perform_now
+      article1 = FactoryBot.create(:page, :published, request_path: "/pwa-showcase")
+      article2 = FactoryBot.create(:page, :published, request_path: "/articles/introducing-joy-of-rails")
+      article1.update_in_search_index
+      article2.update_in_search_index
 
       post search_path, params: {query: "Progressive Web Apps"}
 

spec/system/searches_spec.rb

@@ -2,8 +2,8 @@
 
 RSpec.describe "Searches", type: :system do
   it "show search results" do
-    FactoryBot.create(:page, :published, request_path: "/pwa-showcase")
-    Pages::RefreshSearchIndexJob.perform_now
+    article = FactoryBot.create(:page, :published, request_path: "/pwa-showcase")
+    article.update_in_search_index
 
     visit root_path