FOLIOSYNC-7 Add method to retrieve SRS MARC

README.md

@@ -54,6 +54,9 @@ client.find_instance_record(instance_record_id: 'some-instance-record-id')
 client.find_instance_record(instance_record_hrid: 'some-instance-record-hrid')
 client.find_source_record(instance_record_id: 'some-instance-record-id')
 client.find_source_record(instance_record_hrid: 'some-instance-record-hrid')
+client.find_source_marc_records(modified_since: '2025-01-01T00:00:00Z') { |marc_record_hash| }
+client.find_source_marc_records(with_965_value: '965abc') { |marc_record_hash| }
+client.find_source_marc_records(modified_since: '2025-01-01T00:00:00Z', with_965_value: '965abc') { |marc_record_hash| }
 
 # Convert a FOLIO MARC source record to a marc gem MARC::Record object:
 source_record = client.find_source_record(instance_record_id: 'some-instance-record-id')

lib/folio_api_client/finders.rb

@@ -1,7 +1,7 @@
 # frozen_string_literal: true
 
 class FolioApiClient
-  module Finders
+  module Finders # rubocop:disable Metrics/ModuleLength
     def find_item_record(barcode:)
       item_search_results = self.get('/item-storage/items', { query: "barcode==#{barcode}", limit: 2 })['items']
       return nil if item_search_results.empty?
@@ -90,6 +90,28 @@ def find_source_record(instance_record_id: nil, instance_record_hrid: nil)
       source_record_search_results['sourceRecords'].first
     end
 
+    def find_source_marc_records(modified_since: nil, with_965_value: nil, &block)
+      query = marc_records_query(modified_since: modified_since, with_965_value: with_965_value)
+
+      loop do
+        response = self.get('search/instances', query)
+        process_marc_for_instance(response['instances'], &block) if block
+        break if (query[:offset] + query[:limit]) >= response['totalRecords']
+
+        query[:offset] += query[:limit]
+      end
+    end
+
+    def process_marc_for_instance(instances, &block)
+      instances.each do |instance|
+        source_record = find_source_record(instance_record_id: instance['id'])
+        next if source_record.nil? # Occasionally, we find an instance record without a source record.  Skip these.
+
+        marc_content = source_record.dig('parsedRecord', 'content')
+        yield(marc_content) if marc_content && block
+      end
+    end
+
     def source_record_query(instance_record_id: nil, instance_record_hrid: nil)
       return { instanceId: instance_record_id } if instance_record_id
       return { instanceHrid: instance_record_hrid } if instance_record_hrid
@@ -112,5 +134,26 @@ def location_record_query(code: nil)
       raise FolioApiClient::Exceptions::MissingQueryFieldError,
             'Missing query field.  Must supply a code.'
     end
+
+    def marc_records_query(modified_since: nil, with_965_value: nil) # rubocop:disable Metrics/MethodLength
+      params = { limit: 100, offset: 0 }
+
+      if modified_since.nil? && with_965_value.nil?
+        raise FolioApiClient::Exceptions::MissingQueryFieldError,
+              'Missing query field.  Must supply either modified_since or with_965_value.'
+      end
+
+      if modified_since && !modified_since.match(/^\d{4}-\d{2}-\d{2}T\d{2}:\d{2}:\d{2}Z$/)
+        raise ArgumentError,
+              %(Invalid format for modified_since argument. Must be a date string like "2025-10-03T16:49:00Z".)
+      end
+
+      query_parts = []
+      query_parts << "metadata.updatedDate>=\"#{modified_since}\"" if modified_since
+      query_parts << %(identifiers.value="#{with_965_value}") if with_965_value
+
+      params[:query] = query_parts.join(' and ')
+      params
+    end
   end
 end

spec/folio_api_client/finders_spec.rb

@@ -331,6 +331,80 @@
     end
   end
 
+  describe '#find_source_marc_records' do
+    let(:modified_since) { '2025-01-01T00:00:00Z' }
+    let(:value_965) { '965hyacinth' }
+    let(:marc_content) { { 'fields' => [{ '001' => '12345' }] } }
+    let(:instance_record) { { 'id' => 'instance-123' } }
+    let(:instances_response) { { 'totalRecords' => 1, 'instances' => [instance_record] } }
+
+    before do
+      allow(instance).to receive(:find_source_record).and_return({ 'parsedRecord' => { 'content' => marc_content } })
+    end
+
+    context 'with modified_since parameter' do
+      before do
+        allow(instance).to receive(:get).with(
+          'search/instances', { query: "metadata.updatedDate>=\"#{modified_since}\"", limit: 100, offset: 0 }
+        ).and_return(instances_response)
+      end
+
+      it 'yields MARC content for each source record' do
+        yielded_records = []
+        instance.find_source_marc_records(modified_since: modified_since) { |record| yielded_records << record }
+        expect(yielded_records).to eq([marc_content])
+      end
+
+      it 'raises an exception if the given modified_since parameter is an invalid format' do
+        expect {
+          instance.find_source_marc_records(modified_since: 'banana') { |_| }
+        }.to raise_error(ArgumentError)
+      end
+    end
+
+    context 'with with_965_value parameter' do
+      before do
+        allow(instance).to receive(:get).with(
+          'search/instances', { query: %(identifiers.value="#{value_965}"), limit: 100, offset: 0 }
+        ).and_return(instances_response)
+      end
+
+      it 'uses the correct query' do
+        instance.find_source_marc_records(with_965_value: value_965) { |_| }
+        expect(instance).to have_received(:get).with(
+          'search/instances', { query: 'identifiers.value="965hyacinth"', limit: 100, offset: 0 }
+        )
+      end
+    end
+
+    context 'with modified_since and with_965_value parameters' do
+      before do
+        allow(instance).to receive(:get).with(
+          'search/instances', {
+            query: %(metadata.updatedDate>="#{modified_since}" and identifiers.value="#{value_965}"),
+            limit: 100, offset: 0
+          }
+        ).and_return(instances_response)
+      end
+
+      it 'combines both filters' do
+        instance.find_source_marc_records(modified_since: modified_since, with_965_value: value_965) { |_| }
+        expect(instance).to have_received(:get).with(
+          'search/instances', {
+            query: %(metadata.updatedDate>="#{modified_since}" and identifiers.value="#{value_965}"),
+            limit: 100, offset: 0
+          }
+        )
+      end
+    end
+
+    it 'raises an error when no parameters provided' do
+      expect {
+        instance.find_source_marc_records { |_| }
+      }.to raise_error(FolioApiClient::Exceptions::MissingQueryFieldError)
+    end
+  end
+
   describe '#source_record_query' do
     let(:instance_record_id) { 'instance-record-id' }
     let(:instance_record_hrid) { 'instance-record-hrid' }