Validate that solr doc has title

lib/dpul_collections/indexing_pipeline/figgy/hydration_cache_entry.ex

@@ -34,8 +34,16 @@ defmodule DpulCollections.IndexingPipeline.Figgy.HydrationCacheEntry do
       page_count_i: page_count(metadata),
       image_service_urls_ss: image_service_urls(metadata, related_data)
     }
+    |> validate_solr_document()
   end
 
+  defp validate_solr_document(%{id: id, title_txtm: nil}) do
+    Logger.warning("Solr document is missing title: #{id}")
+    %{}
+  end
+
+  defp validate_solr_document(doc), do: doc
+
   defp image_service_urls(%{"member_ids" => member_ids}, related_data) do
     member_ids
     |> Enum.map(&extract_service_url(&1, related_data))

lib/dpul_collections/solr.ex

@@ -102,7 +102,7 @@ defmodule DpulCollections.Solr do
     response =
       Req.post!(
         update_url(collection),
-        json: docs
+        json: docs |> Enum.filter(fn doc -> map_size(doc) > 0 end)
       )
 
     if response.status != 200 do
@@ -114,7 +114,7 @@ defmodule DpulCollections.Solr do
     response =
       Req.post!(
         update_url(collection),
-        json: docs
+        json: docs |> Enum.filter(fn doc -> map_size(doc) > 0 end)
       )
 
     if response.status != 200 do

test/dpul_collections/indexing_pipeline/figgy/hydration_cache_entry_test.exs

@@ -282,5 +282,43 @@ defmodule DpulCollections.IndexingPipeline.Figgy.HydrationCacheEntryTest do
       assert capture_log(fn -> HydrationCacheEntry.to_solr_document(entry) end) =~
                "couldn't parse date"
     end
+
+    @tag capture_log: true
+    test "an empty solr document is returned with a missing title" do
+      {:ok, entry} =
+        IndexingPipeline.write_hydration_cache_entry(%{
+          cache_version: 0,
+          record_id: "f134f41f-63c5-4fdf-b801-0774e3bc3b2d",
+          source_cache_order: ~U[2018-03-09 20:19:36.465203Z],
+          data: %{
+            "id" => "f134f41f-63c5-4fdf-b801-0774e3bc3b2d",
+            "internal_resource" => "EphemeraFolder",
+            "metadata" => %{
+              "date_created" => ["2022"]
+            }
+          }
+        })
+
+      assert HydrationCacheEntry.to_solr_document(entry) == %{}
+    end
+
+    test "a message is logged when a solr document is missing a title" do
+      {:ok, entry} =
+        IndexingPipeline.write_hydration_cache_entry(%{
+          cache_version: 0,
+          record_id: "f134f41f-63c5-4fdf-b801-0774e3bc3b2d",
+          source_cache_order: ~U[2018-03-09 20:19:36.465203Z],
+          data: %{
+            "id" => "f134f41f-63c5-4fdf-b801-0774e3bc3b2d",
+            "internal_resource" => "EphemeraFolder",
+            "metadata" => %{
+              "date_created" => ["2022"]
+            }
+          }
+        })
+
+      assert capture_log(fn -> HydrationCacheEntry.to_solr_document(entry) end) =~
+               "missing title: f134f41f-63c5-4fdf-b801-0774e3bc3b2d"
+    end
   end
 end

test/dpul_collections/indexing_pipeline/integration/figgy/indexing_integration_test.exs

@@ -115,6 +115,22 @@ defmodule DpulCollections.IndexingPipeline.Figgy.IndexingIntegrationTest do
     indexer |> Broadway.stop(:normal)
   end
 
+  test "when cache entry has an empty document, solr doesn't index it or return an error" do
+    IndexingPipeline.write_transformation_cache_entry(%{
+      cache_version: 0,
+      record_id: "17276197-e223-471c-99d7-405c5f6c5285",
+      source_cache_order: ~U[1999-03-09 20:19:34.486004Z],
+      data: %{}
+    })
+
+    indexer = start_indexing_producer()
+    MockFiggyIndexingProducer.process(1)
+    assert_receive {:ack_done}, 500
+    Solr.commit(active_collection())
+    assert Solr.document_count() == 0
+    indexer |> Broadway.stop(:normal)
+  end
+
   test "solr collection creation" do
     cache_version = 0
     new_collection = "new_index1"

test/dpul_collections/solr_test.exs

@@ -232,4 +232,17 @@ defmodule DpulCollections.SolrTest do
     Solr.commit(active_collection())
     assert Solr.find_by_id(valid_doc["id"])["id"] == valid_doc["id"]
   end
+
+  test "a valid solr document is indexed when in the same batch as an empty document" do
+    valid_doc = %{
+      "id" => "e0602353-4429-4405-b080-064952f9b267",
+      "title_txtm" => ["test title 1"]
+    }
+
+    empty_doc = %{}
+
+    Solr.add([valid_doc, empty_doc], active_collection())
+    Solr.commit(active_collection())
+    assert Solr.find_by_id(valid_doc["id"])["id"] == valid_doc["id"]
+  end
 end