From 7b60d407854883e48b740e52bcd60961fc580b19 Mon Sep 17 00:00:00 2001
From: jrhoads <jrhoads@datacite.org>
Date: Tue, 14 Feb 2023 13:31:58 -0500
Subject: [PATCH] Add rake task to back-fill disciplinary clients/repositories
 that have re3data subjects.

---
 lib/tasks/enrich.rake | 29 +++++++++++++++++++++++++++++
 1 file changed, 29 insertions(+)
 create mode 100644 lib/tasks/enrich.rake

diff --git a/lib/tasks/enrich.rake b/lib/tasks/enrich.rake
new file mode 100644
index 000000000..91ec75f4e
--- /dev/null
+++ b/lib/tasks/enrich.rake
@@ -0,0 +1,29 @@
+# frozen_string_literal: true
+
+namespace :enrich do
+  desc "Enrich Clients with Subjects from re3data and converted to Field Of Science subjectScheme"
+  task client_subjects: :environment do
+    def enrich_client(client)
+      re3data = DataCatalog.find_by_id(client.re3data_id).fetch(:data, []).first
+      if re3data
+        subs = re3data.subjects
+        dfg_ids = subs.select { |subject|
+          subject.scheme == "DFG"
+        }.map { |subject|
+          subject.text.split.first
+        }
+        fos_subjects = Bolognese::Utils.dfg_ids_to_fos(dfg_ids)
+        client.subjects = fos_subjects.uniq
+        client.save
+      end
+    end
+
+    puts "Searching for disciplinary repositories with re3data_ids without subjects"
+    search_results = Client.search("re3data_id:* AND -subjects:* AND repository_type:disciplinary")
+    puts "Found #{search_results.records.count} repostitories.  Enriching now..."
+    search_results.records.map do |c|
+      enrich_client(c)
+    end
+    puts "Enrichment complete"
+  end
+end