BL-1742 Add extract_mac_subfield_limit

lib/cob_index/indexer_config.rb

@@ -87,21 +87,22 @@
   711fgklnpst
 
                                              }.join(":")), wrap_begin_end
+
+to_field "title_added_entry_authority_id_ms", extract_marc_subfield_limit("7000:7100:7110", "t", true)
+to_field "title_added_entry_real_world_object_uri_ms", extract_marc_subfield_limit("7001:7101:7111", "t", true)
+
 to_field "title_sort", extract_marc("245abcfgknps", alternate_script: false, first: true)
 
 # Creator/contributor fields
 to_field "creator_t", extract_marc("245c:100abcdejlmnopqrtu:110abcdelmnopt:111acdejlnopt:700abcdejqu:710abcde:711acdej", trim_punctuation: true), delete_if(CORPORATE_NAMES), wrap_begin_end
 to_field "creator_authority_record_id_ms", extract_marc("1000:1100:1110")
 to_field "creator_real_world_object_uri_ms", extract_marc("1001:1101:1111")
 
-
-
 to_field "creator_facet", extract_marc("100abcdq:110abcd:111ancdj:700abcdq:710abcd:711ancdj", trim_punctuation: true), delete_if(CORPORATE_NAMES)
 to_field "creator_display", extract_creator, delete_if(CORPORATE_NAMES)
 to_field "contributor_display", extract_contributor, delete_if(Proc.new { |v| CORPORATE_NAMES.include?(JSON.parse(v)["name"]) })
-to_field "contributor_authority_record_id_ms", extract_marc("7000:7100:7110")
-to_field "contributor_real_world_object_uri_ms", extract_marc("7001:7101:7111")
-to_field "title_added_entry_authority_id_ms", extract_marc("700t:710t:711t")
+to_field "contributor_authority_record_id_ms", extract_marc_subfield_limit("7000:7100:7110", "t", false)
+to_field "contributor_real_world_object_uri_ms", extract_marc_subfield_limit("7001:7101:7111", "t", false)
 
 to_field "creator_vern_display", extract_creator_vern, delete_if(CORPORATE_NAMES)
 to_field "contributor_vern_display", extract_contributor_vern, delete_if(CORPORATE_NAMES)

lib/cob_index/macros/custom.rb

@@ -890,4 +890,30 @@ def extract_donor
       end
     end
   end
+
+  def extract_marc_subfield_limit(spec, subfield_limit, boolean)
+    # spec is the standard extract_marc string.
+    # subfield_limit is a string of the subfield code used to limit the extracted marc output.
+    # boolean is either true or false.
+    # If true, extract_marc_subfield_limit will only extract a field if that field also includes a subfield that matches the subfield_limit code.
+    # If false, extract_marc_subfield_limit will only extract a field from spec if it does not include a subfield that matches the subfield_limit code.
+
+    spec_array = spec.split(":")
+
+    lambda do |rec, acc|
+      spec_array.each do |spec_subset|
+        tag = spec_subset[0, 3]
+        if (boolean == true && rec.fields(tag).any? { |field| field[subfield_limit].present? }) ||
+        (boolean == false && rec.fields(tag).none? { |field| field[subfield_limit].present? })
+          values = Traject::MarcExtractor.cached(spec_subset).collect_matching_lines(rec) do |field, spec_subset, extractor|
+            extractor.collect_subfields(field, spec_subset)
+          end.compact
+          acc.concat(values)
+        else
+          acc
+        end
+
+      end
+    end
+  end
 end

spec/cob_index/macros/custom_spec.rb

@@ -2517,4 +2517,72 @@
       end
     end
   end
+
+  describe "#extract_marc_subfield_limit for title_added_entry_authority_id_ms" do
+
+    let(:path) { "extract_marc_subfield_limit.xml" }
+
+    before do
+      subject.instance_eval do
+        to_field "title_added_entry_authority_id_ms", extract_marc_subfield_limit("7000:7100:7110", "t", true)
+
+        settings do
+          provide "marc_source.type", "xml"
+        end
+      end
+    end
+
+    context "no field" do
+      it "does not error out" do
+        expect(subject.map_record(records[0])).to eq({})
+      end
+    end
+
+    context "subtitle limit matches" do
+      it "maps field" do
+        expect(subject.map_record(records[1])).to eq("title_added_entry_authority_id_ms" => ["https://id.loc.gov/authorities/names/no95021615"])
+      end
+    end
+
+    context "subtitle limit does not match" do
+      it "does not map field" do
+        expect(subject.map_record(records[2])).to eq({})
+      end
+    end
+
+  end
+
+  describe "#extract_marc_subfield_limit for contributor_authority_record_id_ms" do
+
+    let(:path) { "extract_marc_subfield_limit.xml" }
+
+    before do
+      subject.instance_eval do
+        to_field "contributor_authority_record_id_ms", extract_marc_subfield_limit("7000:7100:7110", "t", false)
+
+        settings do
+          provide "marc_source.type", "xml"
+        end
+      end
+    end
+
+    context "no field" do
+      it "does not error out" do
+        expect(subject.map_record(records[0])).to eq({})
+      end
+    end
+
+    context "subtitle limit matches" do
+      it "does not map field" do
+        expect(subject.map_record(records[1])).to eq({})
+      end
+    end
+
+    context "subtitle limit does not match" do
+      it "maps record" do
+        expect(subject.map_record(records[2])).to eq("contributor_authority_record_id_ms" => ["https://id.loc.gov/authorities/names/no2017022085"])
+      end
+    end
+
+  end
 end

spec/fixtures/marc_files/extract_marc_subfield_limit.xml

@@ -0,0 +1,24 @@
+<?xml version="1.0" encoding="UTF-8"?>
+<collection xmlns='http://www.loc.gov/MARC21/slim' xmlns:xsi='http://www.w3.org/2001/XMLSchema-instance' xsi:schemaLocation='http://www.loc.gov/MARC21/slim http://www.loc.gov/standards/marcxml/schema/MARC21slim.xsd'>
+<record>
+</record>
+<record>
+    <datafield ind1="1" ind2="2" tag="700">
+        <subfield code="a">Mozart, Wolfgang Amadeus,</subfield>
+        <subfield code="d">1756-1791.</subfield>
+        <subfield code="t">Symphonies,</subfield>
+        <subfield code="n">K. 73,</subfield>
+        <subfield code="r">C major</subfield>
+        <subfield code="0">https://id.loc.gov/authorities/names/no95021615</subfield>
+    </datafield>
+</record>
+<record>
+    <datafield ind1="1" ind2=" " tag="700">
+        <subfield code="a">Jenkins, Barry,</subfield>
+        <subfield code="d">1979-</subfield>
+        <subfield code="0">https://id.loc.gov/authorities/names/no2017022085</subfield>
+        <subfield code="e">film director,</subfield>
+        <subfield code="e">screenwriter.</subfield>
+    </datafield>
+</record>
+</collection>