Merge pull request #8121 from alphagov/delete-search-index-records-wh…

…en-edition-becomes-unindexable

Delete search index records when an edition becomes unindexable

app/services/service_listeners/search_indexer.rb

@@ -8,6 +8,10 @@ def index!
 
         Whitehall::SearchIndex.add(edition)
         reindex_collection_documents
+      elsif edition.previous_edition&.can_index_in_search?
+        # If the previous edition was indexed but the current edition cannot be, we must delete the previous edition from the index
+        Whitehall::SearchIndex.delete(edition.previous_edition)
+        reindex_collection_documents
       end
     end
 

db/data_migration/20230816152924_remove_orphaned_edition_search_index_records.rb

@@ -0,0 +1,9 @@
+class RemoveOrphanedEditionSearchIndexRecords < ActiveRecord::Migration[7.0]
+  def change
+    # We want to find all editions which are publicly visible and therefore might once have been indexed, but are currently not searchable
+    # We then ask Search API to remove the record for these editions if a corresponding record is present
+    # Note that this doesn't seem to capture all foreign language content in the search index, but it has removed at least some
+    editions_with_possible_orphaned_search_index_records = Edition.publicly_visible.where.not(id: Edition.search_only.select(:id))
+    editions_with_possible_orphaned_search_index_records.each { |e| ServiceListeners::SearchIndexer.new(e).remove! }
+  end
+end

test/unit/app/services/service_listeners/search_indexer_test.rb

@@ -41,6 +41,17 @@ class ServiceListeners::SearchIndexerTest < ActiveSupport::TestCase
     ServiceListeners::SearchIndexer.new(edition).index!
   end
 
+  test "#index! removes the edition from the search index if the current edition cannot be indexed, but the previous edition was indexed" do
+    english_edition = create(:news_article_world_news_story, :published)
+    ServiceListeners::SearchIndexer.new(english_edition).index!
+    non_english_edition = I18n.with_locale(:fr) do
+      create(:news_article_world_news_story, :published, primary_locale: :fr, document: english_edition.document)
+    end
+
+    expect_removal_from_index(english_edition)
+    ServiceListeners::SearchIndexer.new(non_english_edition).index!
+  end
+
   test "#remove! removes the edition from the search index" do
     edition = create(:published_news_article)