From f28c8cabc797e01585704e8c5ee077e7da2ce7a6 Mon Sep 17 00:00:00 2001
From: Eliot Jordan <eliotj@princeton.edu>
Date: Mon, 25 Nov 2024 12:34:05 -0600
Subject: [PATCH] Assign default when title is missing

---
 .../figgy/hydration_cache_entry.ex            | 10 ++++++++--
 lib/dpul_collections/solr.ex                  |  8 +++++++-
 .../figgy/hydration_cache_entry_test.exs      | 19 +++++++++++++++++++
 .../figgy/indexing_integration_test.exs       | 16 ++++++++++++++++
 test/dpul_collections/solr_test.exs           |  2 --
 5 files changed, 50 insertions(+), 5 deletions(-)

diff --git a/lib/dpul_collections/indexing_pipeline/figgy/hydration_cache_entry.ex b/lib/dpul_collections/indexing_pipeline/figgy/hydration_cache_entry.ex
index 8628c88e..6b211663 100644
--- a/lib/dpul_collections/indexing_pipeline/figgy/hydration_cache_entry.ex
+++ b/lib/dpul_collections/indexing_pipeline/figgy/hydration_cache_entry.ex
@@ -27,7 +27,7 @@ defmodule DpulCollections.IndexingPipeline.Figgy.HydrationCacheEntry do
 
     %{
       id: id,
-      title_txtm: get_in(metadata, ["title"]),
+      title_txtm: extract_title(metadata),
       description_txtm: get_in(metadata, ["description"]),
       years_is: extract_years(data),
       display_date_s: format_date(metadata),
@@ -49,6 +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",
@@ -82,7 +84,11 @@ defmodule DpulCollections.IndexingPipeline.Figgy.HydrationCacheEntry do
 
   defp extract_service_url(nil), do: nil
 
-  defp extract_service_url(_id, _), do: nil
+  def extract_title(%{"title" => []}) do
+    ["[Missing Title]"]
+  end
+
+  def extract_title(%{"title" => title}), do: title
 
   defp is_derivative(%{
          "mime_type" => ["image/tiff"],
diff --git a/lib/dpul_collections/solr.ex b/lib/dpul_collections/solr.ex
index 6ad1695c..33f4f333 100644
--- a/lib/dpul_collections/solr.ex
+++ b/lib/dpul_collections/solr.ex
@@ -98,7 +98,9 @@ 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),
@@ -108,6 +110,8 @@ defmodule DpulCollections.Solr do
     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
@@ -121,6 +125,8 @@ defmodule DpulCollections.Solr 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()}
diff --git a/test/dpul_collections/indexing_pipeline/figgy/hydration_cache_entry_test.exs b/test/dpul_collections/indexing_pipeline/figgy/hydration_cache_entry_test.exs
index 0d5c0d32..9f5d333f 100644
--- a/test/dpul_collections/indexing_pipeline/figgy/hydration_cache_entry_test.exs
+++ b/test/dpul_collections/indexing_pipeline/figgy/hydration_cache_entry_test.exs
@@ -282,5 +282,24 @@ defmodule DpulCollections.IndexingPipeline.Figgy.HydrationCacheEntryTest do
       assert capture_log(fn -> HydrationCacheEntry.to_solr_document(entry) end) =~
                "couldn't parse date"
     end
+
+    test "an empty solr document is returned with a empty title field" 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" => %{
+              "title" => [],
+              "date_created" => ["2022"]
+            }
+          }
+        })
+
+      assert %{title_txtm: ["[Missing Title]"]} = HydrationCacheEntry.to_solr_document(entry)
+    end
   end
 end
diff --git a/test/dpul_collections/indexing_pipeline/integration/figgy/indexing_integration_test.exs b/test/dpul_collections/indexing_pipeline/integration/figgy/indexing_integration_test.exs
index ac47edc5..5963ae3f 100644
--- a/test/dpul_collections/indexing_pipeline/integration/figgy/indexing_integration_test.exs
+++ b/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"
diff --git a/test/dpul_collections/solr_test.exs b/test/dpul_collections/solr_test.exs
index 9b9c91f8..912371ef 100644
--- a/test/dpul_collections/solr_test.exs
+++ b/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