Merge pull request #5124 from sul-dlss/metadata_transfer_tweaks

app/services/publish/metadata_transfer_service.rb

@@ -140,7 +140,10 @@ def workspace_content_pathname
     def copy(src_pathname, dest_pathname)
       return if dest_pathname.exist?
 
+      Rails.logger.info("Copying #{src_pathname} (#{src_pathname.size} bytes) to #{dest_pathname} for #{druid}")
       FileUtils.copy(src_pathname, dest_pathname)
+
+      raise "Copy #{src_pathname} to #{dest_pathname} failed" unless dest_pathname.exist? && dest_pathname.size == src_pathname.size
     end
 
     def transfer_stage_pathname

app/services/shelvable_files_stager.rb

@@ -39,15 +39,26 @@ def stage
 
   # Try to copy from preservation into the workspace
   def copy_file_from_preservation(file_pathname:, filepath:)
+    Rails.logger.info("Copying #{filepath} from preservation to #{file_pathname} for #{druid}")
     FileUtils.mkdir_p(file_pathname.dirname)
+    received_bytes = 0
     File.open(file_pathname, 'wb') do |streamed|
-      writer = proc do |chunk, _overall_received_bytes|
+      writer = proc do |chunk, overall_received_bytes|
         streamed.write chunk
+        received_bytes = overall_received_bytes
       end
-      Preservation::Client.objects.content(druid:, filepath:, version: version - 1, on_data: writer)
+      response = Preservation::Client.objects.content(druid:, filepath:, version: version - 1, on_data: writer)
+      check_filesize(file_pathname:, filepath:, expected: response&.headers&.[]('content-length')&.to_i, received: received_bytes)
     end
   rescue Preservation::Client::NotFoundError, Faraday::ResourceNotFound
     file_pathname.delete if file_pathname.exist? # 404 body is written to file
     raise FileNotFound, "Unable to find #{filepath} in the content directory"
   end
+
+  def check_filesize(file_pathname:, filepath:, expected:, received:)
+    return if expected.nil? || expected == received
+
+    file_pathname.delete if file_pathname.exist?
+    raise "File copied from preservation was not the expected size. Expected #{expected} bytes for #{filepath}; received #{received} bytes."
+  end
 end

spec/services/shelvable_files_stager_spec.rb

@@ -55,4 +55,14 @@
       expect { stage }.to raise_error(ShelvableFilesStager::FileNotFound)
     end
   end
+
+  context 'when the content length does not match' do
+    before do
+      allow(Preservation::Client.objects).to receive(:content).and_return(instance_double(Faraday::Response, headers: { 'content-length' => '42' }))
+    end
+
+    it 'raises' do
+      expect { stage }.to raise_error('File copied from preservation was not the expected size. Expected 42 bytes for file1.txt; received 0 bytes.')
+    end
+  end
 end