AO3-7284 Configurable indexer batch size

app/models/search/indexer.rb

@@ -1,5 +1,4 @@
 class Indexer
-  BATCH_SIZE = 1000
   INDEXERS_FOR_CLASS = {
     Work: %w[WorkIndexer WorkCreatorIndexer BookmarkedWorkIndexer],
     Bookmark: %w[BookmarkIndexer],
@@ -124,15 +123,19 @@ def self.index_all(options = {})
   end
 
   def self.index_from_db
-    total = (indexables.count / BATCH_SIZE) + 1
+    total = (indexables.count / batch_size) + 1
     i = 1
-    indexables.find_in_batches(batch_size: BATCH_SIZE) do |group|
+    indexables.find_in_batches(batch_size: batch_size) do |group|
       puts "Queueing #{klass} batch #{i} of #{total}" unless Rails.env.test?
       AsyncIndexer.new(self, :world).enqueue_ids(group.map(&:id))
       i += 1
     end
   end
 
+  def self.batch_size
+    ArchiveConfig.SEARCH_INDEXER_BATCH_SIZE
+  end
+
   # Add conditions here
   def self.indexables
     klass.constantize

config/config.yml

@@ -706,6 +706,9 @@ MINUTES_UNTIL_COLLECTION_BLURBS_EXPIRE: 120
 # Job size used for the HitCountUpdateJob class:
 HIT_COUNT_JOB_SIZE: 1000
 
+# Batch size used by Indexer when reading records from the database:
+SEARCH_INDEXER_BATCH_SIZE: 1000
+
 # Batch size used for the HitCountUpdateJob class:
 HIT_COUNT_BATCH_SIZE: 100
 

spec/models/search/indexer_spec.rb

@@ -0,0 +1,20 @@
+require "spec_helper"
+
+describe Indexer do
+  describe ".index_from_db" do
+    let!(:work1) { create(:work) }
+    let!(:work2) { create(:work) }
+    let(:async_indexer) { instance_double(AsyncIndexer, enqueue_ids: true) }
+
+    before do
+      allow(ArchiveConfig).to receive(:SEARCH_INDEXER_BATCH_SIZE).and_return(1)
+      allow(AsyncIndexer).to receive(:new).and_return(async_indexer)
+    end
+
+    it "uses the configured batch size when indexing" do
+      WorkIndexer.index_from_db
+
+      expect(async_indexer).to have_received(:enqueue_ids).twice
+    end
+  end
+end