Pull year values from different types of date strings

lib/dpul_collections/indexing_pipeline/figgy/hydration_cache_entry.ex

@@ -114,15 +114,13 @@ defmodule DpulCollections.IndexingPipeline.Figgy.HydrationCacheEntry do
   # If somehow we get more than 1 value, just take the first
   # It goes into a multi-valued index field, so keep it looking that way
   defp extract_years(%{"metadata" => %{"date_created" => [date | _tail]}, "id" => id}) do
-    result = Integer.parse(date)
-
-    case result do
-      :error ->
+    case parse_date(date) do
+      nil ->
         Logger.warning("couldn't parse date \"#{date}\" for record #{id}")
         nil
 
-      {extracted_date, _rest} ->
-        [extracted_date]
+      x ->
+        [String.to_integer(x)]
     end
   end
 
@@ -142,7 +140,7 @@ defmodule DpulCollections.IndexingPipeline.Figgy.HydrationCacheEntry do
   end
 
   defp format_date(%{"date_created" => [date | _tail]}) do
-    date
+    parse_date(date)
   end
 
   defp format_date(%{"date_created" => []}) do
@@ -153,4 +151,16 @@ defmodule DpulCollections.IndexingPipeline.Figgy.HydrationCacheEntry do
     # there's no date_created value
     nil
   end
+
+  defp parse_date(date_string) do
+    [
+      # "29 Raḥab al-Marjab 1321- [July 1923]"
+      ~r/\[*(\d+)(?=\])/,
+      # "November 1952" or "1943"
+      ~r/\d{4}/
+    ]
+    |> Enum.map(fn regex -> Regex.run(regex, date_string, capture: :first) end)
+    |> Enum.find([nil], fn s -> s != nil end)
+    |> hd
+  end
 end

test/dpul_collections/indexing_pipeline/figgy/hydration_cache_entry_test.exs

@@ -130,25 +130,82 @@ defmodule DpulCollections.IndexingPipeline.Figgy.HydrationCacheEntryTest do
           }
         })
 
+      # Add one to exercise date_created with format `.*yyyy`
+      {:ok, entry6} =
+        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" => ["test title 6"],
+              "date_created" => ["January 26, 1952"]
+            }
+          }
+        })
+
+      # Add one to exercise date_created with format `.*[.*yyyy]`
+      {:ok, entry7} =
+        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" => ["test title 7"],
+              "date_created" => ["29 Raḥab al-Marjab 1342- رحب المرجب 1342 - [July 1923]"]
+            }
+          }
+        })
+
+      # Add one to exercise date_created with format `.*[.*yyyy]`
+      {:ok, entry8} =
+        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" => ["test title 8"],
+              "date_created" => ["September 1931"]
+            }
+          }
+        })
+
       doc4 = HydrationCacheEntry.to_solr_document(entry4)
       doc5 = HydrationCacheEntry.to_solr_document(entry5)
+      doc6 = HydrationCacheEntry.to_solr_document(entry6)
+      doc7 = HydrationCacheEntry.to_solr_document(entry7)
+      doc8 = HydrationCacheEntry.to_solr_document(entry8)
 
       assert doc1[:years_is] == [2022]
       assert doc2[:years_is] == [1995, 1996, 1997, 1998, 1999, 2000, 2001, 2002, 2003, 2004, 2005]
       assert doc3[:years_is] == nil
       assert doc4[:years_is] == [2011, 2012, 2013]
       assert doc5[:years_is] == nil
+      assert doc6[:years_is] == [1952]
+      assert doc7[:years_is] == [1923]
+      assert doc8[:years_is] == [1931]
 
       assert doc1[:display_date_s] == "2022"
       assert doc2[:display_date_s] == "1995 - 2005"
       assert doc3[:display_date_s] == nil
       assert doc4[:display_date_s] == "2011 - 2013 (approximate)"
       assert doc5[:display_date_s] == nil
+      assert doc6[:display_date_s] == "1952"
+      assert doc7[:display_date_s] == "1923"
+      assert doc8[:display_date_s] == "1931"
     end
 
     test "logs dates it can't parse" do
       # date created has a bad date
-      {:ok, entry6} =
+      {:ok, entry} =
         IndexingPipeline.write_hydration_cache_entry(%{
           cache_version: 0,
           record_id: "f134f41f-63c5-4fdf-b801-0774e3bc3b2d",
@@ -163,8 +220,7 @@ defmodule DpulCollections.IndexingPipeline.Figgy.HydrationCacheEntryTest do
           }
         })
 
-      # doc6 = HydrationCacheEntry.to_solr_document(entry6)
-      assert capture_log(fn -> HydrationCacheEntry.to_solr_document(entry6) end) =~
+      assert capture_log(fn -> HydrationCacheEntry.to_solr_document(entry) end) =~
                "couldn't parse date"
     end
   end