From d8b0b62f6cd83a63fd9c85ecdf615ab1391ca415 Mon Sep 17 00:00:00 2001
From: Kirk Wang <k3wang@gmail.com>
Date: Mon, 19 Feb 2024 05:46:13 -0800
Subject: [PATCH] Add conditional to `.fileset_for_directives` (#6695)

We are noticing that when ingesting a PDF, the
`ValkyrieCreateDerivativesJob` is failing because the file_set doesn't
get found in `ValkyriePersistDerivatives.fileset_for_directives` when
the directives[:url] is an id instead of an actual uri, see
`FileSetDerivativesService#extract_full_text`.  I'm suggesting that we
check if we have a "/" and if not we assume that it is an id so we can
just look for that.
---
 app/services/hyrax/valkyrie_persist_derivatives.rb  | 13 ++++++++++---
 .../hyrax/valkyrie_persist_derivatives_spec.rb      | 12 ++++++++++++
 2 files changed, 22 insertions(+), 3 deletions(-)

diff --git a/app/services/hyrax/valkyrie_persist_derivatives.rb b/app/services/hyrax/valkyrie_persist_derivatives.rb
index 910f5b24a3..78bccf322b 100644
--- a/app/services/hyrax/valkyrie_persist_derivatives.rb
+++ b/app/services/hyrax/valkyrie_persist_derivatives.rb
@@ -43,9 +43,16 @@ def self.call(stream,
     # @return [Hyrax::FileSet]
     def self.fileset_for_directives(directives)
       path = URI(directives.fetch(:url)).path
-      id = path.sub(Hyrax.config.derivatives_path.to_s, "")
-               .delete('/')
-               .match(/^(.*)-\w*(\.\w+)*$/) { |m| m[1] }
+      # checks if it's a file path, else assuming it is already an id
+      # Hyrax::FileSetDerivativesService#extract_full_text passes in the raw uri
+      # and not a derivative_url like the other derivative formats
+      id = if path.include?("/")
+             path.sub(Hyrax.config.derivatives_path.to_s, "")
+                 .delete('/')
+                 .match(/^(.*)-\w*(\.\w+)*$/) { |m| m[1] }
+           else
+             path
+           end
       raise "Could not extract fileset id from path #{path}" unless id
 
       Hyrax.metadata_adapter.query_service.find_by(id: id)
diff --git a/spec/services/hyrax/valkyrie_persist_derivatives_spec.rb b/spec/services/hyrax/valkyrie_persist_derivatives_spec.rb
index f91ff013c8..6973d5c868 100644
--- a/spec/services/hyrax/valkyrie_persist_derivatives_spec.rb
+++ b/spec/services/hyrax/valkyrie_persist_derivatives_spec.rb
@@ -90,5 +90,17 @@
           .to raise_error(/Could not extract fileset id from path/)
       end
     end
+
+    context 'with an id' do
+      let(:directives) do
+        { url: '123' }
+      end
+
+      it 'extracts the id' do
+        expect(Hyrax.metadata_adapter.query_service)
+          .to receive(:find_by).with(id: '123')
+        described_class.fileset_for_directives(directives)
+      end
+    end
   end
 end