Need to strip oclc leading zeros for hathi data lookup (#229)

psu-libraries · Aug 4, 2020 · 2f42d91 · 2f42d91
1 parent a77b705
commit 2f42d91
Show file tree

Hide file tree

Showing 3 changed files with 10 additions and 2 deletions.
diff --git a/.rubocop_todo.yml b/.rubocop_todo.yml
@@ -15,7 +15,7 @@ Metrics/AbcSize:
 # Configuration parameters: CountComments, ExcludedMethods.
 # ExcludedMethods: refine
 Metrics/BlockLength:
-  Max: 212
+  Max: 218
 
 # Offense count: 2
 # Configuration parameters: CountComments.

diff --git a/lib/traject/macros/custom.rb b/lib/traject/macros/custom.rb
@@ -207,7 +207,7 @@ def self.includes_oclc_indicators?(sf_a)
 
       def extract_hathi_data
         lambda do |_record, accumulator, context|
-          oclc_number = context.output_hash&.dig('oclc_number_ssim')&.first
+          oclc_number = context.output_hash&.dig('oclc_number_ssim')&.first.to_i.to_s
           ht_hash = HATHI_MULTI_OVERLAP&.dig(oclc_number)&.first || HATHI_MONO_OVERLAP&.dig(oclc_number)&.first
           accumulator << ht_hash.to_json unless ht_hash.nil?
         end

diff --git a/spec/lib/traject/macros/custom_spec.rb b/spec/lib/traject/macros/custom_spec.rb
@@ -218,6 +218,14 @@
         expect(result['hathitrust_struct']).to match ['{"ht_bib_key":"012292266","access":"allow"}']
       end
     end
+
+    context 'when a record has an oclc with leading zeros' do
+      let(:oclc) { { '035' => { 'ind1' => '', 'ind2' => '', 'subfields' => [{ 'a' => '(OCLC)00100000391' }] } } }
+
+      it 'finds a match after stripping the leading zeros' do
+        expect(result['hathitrust_struct']).to match ['{"ht_bib_key":"012292266","access":"allow"}']
+      end
+    end
   end
 
   describe '#hathi_to_hash' do