FIX: Skip deleted or empty content when backfilling translations

app/jobs/scheduled/automatic_translation_backfill.rb

@@ -18,7 +18,12 @@ def execute(args = nil)
       end
     end
 
-    def fetch_untranslated_model_ids(model = Post, limit = 100, target_locales = backfill_locales)
+    def fetch_untranslated_model_ids(
+      model,
+      content_column,
+      limit,
+      target_locales = backfill_locales
+    )
       m = model.name.downcase
       DB.query_single(<<~SQL, target_locales: target_locales, limit: limit)
         SELECT m.id
@@ -39,6 +44,8 @@ def fetch_untranslated_model_ids(model = Post, limit = 100, target_locales = bac
               ARRAY[]::text[]
             ))
         )
+        AND m.deleted_at IS NULL
+        AND m.#{content_column} != ''
         ORDER BY m.id DESC
         LIMIT :limit
       SQL
@@ -92,9 +99,9 @@ def translate_records(type, record_ids)
     def process_batch
       models_translated = [Post, Topic].size
       translations_per_model = [translations_per_run / models_translated, 1].max
-      topic_ids = fetch_untranslated_model_ids(Topic, translations_per_model)
+      topic_ids = fetch_untranslated_model_ids(Topic, "title", translations_per_model)
       translations_per_model = translations_per_run - topic_ids.size
-      post_ids = fetch_untranslated_model_ids(Post, translations_per_model)
+      post_ids = fetch_untranslated_model_ids(Post, "cooked", translations_per_model)
       return if topic_ids.empty? && post_ids.empty?
 
       translate_records(Topic, topic_ids)

spec/jobs/automatic_translation_backfill_spec.rb

@@ -177,8 +177,21 @@ def expect_google_translate(text)
     end
 
     it "returns correct post ids needing translation in descending id" do
-      result = described_class.new.fetch_untranslated_model_ids(Post, 50, %w[de es])
+      result = described_class.new.fetch_untranslated_model_ids(Post, "cooked", 50, %w[de es])
       expect(result).to include(post_7.id, post_6.id, post_3.id, post_2.id, post_1.id)
     end
+
+    it "does not return posts that are deleted" do
+      post_1.trash!
+      result = described_class.new.fetch_untranslated_model_ids(Post, "cooked", 50, %w[de es])
+      expect(result).not_to include(post_1.id)
+    end
+
+    it "does not return posts that are empty" do
+      post_1.cooked = ""
+      post_1.save!(validate: false)
+      result = described_class.new.fetch_untranslated_model_ids(Post, "cooked", 50, %w[de es])
+      expect(result).not_to include(post_1.id)
+    end
   end
 end