Assign default when title is missing

lib/dpul_collections/indexing_pipeline/figgy/hydration_cache_entry.ex

@@ -27,26 +27,15 @@ defmodule DpulCollections.IndexingPipeline.Figgy.HydrationCacheEntry do
 
     %{
       id: id,
-      title_txtm: get_in(metadata, ["title"]),
+      title_txtm: get_in(metadata, ["title"]) || ["Missing Title"],
       description_txtm: get_in(metadata, ["description"]),
       years_is: extract_years(data),
       display_date_s: format_date(metadata),
       page_count_i: page_count(metadata),
       image_service_urls_ss: image_service_urls(metadata, related_data)
     }
-    # Remove k,v with empty values
-    |> Enum.filter(fn {k, v} -> v end)
-    |> Map.new()
-    |> validate_solr_document()
   end
 
-  defp validate_solr_document(doc) when not is_map_key(doc, :title_txtm) do
-    Logger.warning("Solr document is missing title: #{doc[: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))
@@ -60,6 +49,8 @@ defmodule DpulCollections.IndexingPipeline.Figgy.HydrationCacheEntry do
     extract_service_url(member_data[id])
   end
 
+  defp extract_service_url(_id, _), do: nil
+
   # Find the derivative FileMetadata
   defp extract_service_url(%{
          "internal_resource" => "FileSet",
@@ -93,8 +84,6 @@ defmodule DpulCollections.IndexingPipeline.Figgy.HydrationCacheEntry do
 
   defp extract_service_url(nil), do: nil
 
-  defp extract_service_url(_id, _), do: nil
-
   defp is_derivative(%{
          "mime_type" => ["image/tiff"],
          "use" => [%{"@id" => "http://pcdm.org/use#ServiceFile"}]

lib/dpul_collections/solr.ex

@@ -98,29 +98,35 @@ defmodule DpulCollections.Solr do
   end
 
   @spec add(list(map()), String.t()) :: {:ok, Req.Response.t()} | {:error, Exception.t()}
-  def add(docs, collection \\ read_collection()) when length(docs) > 1 do
+  def add(docs, collection \\ read_collection())
+
+  def add(docs, collection) when length(docs) > 1 do
     response =
       Req.post!(
         update_url(collection),
-        json: docs |> Enum.filter(fn doc -> map_size(doc) > 0 end)
+        json: docs
       )
 
     if response.status != 200 do
       Enum.each(docs, fn doc -> add([doc]) end)
     end
+
+    response
   end
 
   def add(docs, collection) when length(docs) when length(docs) == 1 do
     response =
       Req.post!(
         update_url(collection),
-        json: docs |> Enum.filter(fn doc -> map_size(doc) > 0 end)
+        json: docs
       )
 
     if response.status != 200 do
       doc = docs |> Enum.at(0)
       Logger.warning("error indexing solr document with id: #{doc["id"]} #{response.body}")
     end
+
+    response
   end
 
   @spec commit(String.t()) :: {:ok, Req.Response.t()} | {:error, Exception.t()}

test/dpul_collections/indexing_pipeline/figgy/hydration_cache_entry_test.exs

@@ -283,7 +283,6 @@ defmodule DpulCollections.IndexingPipeline.Figgy.HydrationCacheEntryTest do
                "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(%{
@@ -299,26 +298,7 @@ defmodule DpulCollections.IndexingPipeline.Figgy.HydrationCacheEntryTest do
           }
         })
 
-      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"
+      assert %{title_txtm: ["Missing Title"]} = HydrationCacheEntry.to_solr_document(entry)
     end
   end
 end

test/dpul_collections/solr_test.exs

@@ -209,8 +209,6 @@ defmodule DpulCollections.SolrTest do
       "id" => "3cb7627b-defc-401b-9959-42ebc4488f74"
     }
 
-    # Solr.commit(active_collection())
-
     assert capture_log(fn -> Solr.add([doc], active_collection()) end) =~
              "error indexing solr document"
   end
@@ -232,17 +230,4 @@ 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