Merge pull request #733 from pulibrary/i678-keywords

Add keyword configuration option to OCLC newly added

app/models/oclc/lc_call_slips/keyword_field.rb

@@ -0,0 +1,44 @@
+# frozen_string_literal: true
+module Oclc
+  module LcCallSlips
+    # This class is responsible for searching a given
+    # Marc::DataField for the keywords that a selector
+    # is interested in.
+    class KeywordField
+      def initialize(field:, keywords:)
+        @field = field
+        @keywords = keywords
+      end
+
+      def match?
+        keyword_field? && field_contains_keywords?
+      end
+
+      private
+
+      attr_reader :field, :keywords
+
+      def keyword_field?
+        field.is_a?(MARC::DataField) && field.tag.match?(/^[12578]/)
+      end
+
+      def field_contains_keywords?
+        field.any? { |subfield| subfield_contains_keywords?(subfield) }
+      end
+
+      def subfield_contains_keywords?(subfield)
+        words_in_subfield = subfield.value.split(' ')
+        words_in_subfield.any? { |found_word| word_is_keyword? found_word }
+      end
+
+      def word_is_keyword?(word)
+        keywords.any? do |desired_keyword|
+          # Add ^ and $ to make sure that we match the whole world,
+          # then turn the * wildcard into .*
+          desired_keyword_as_regexp = Regexp.new('^' + desired_keyword.gsub('*', '.*') + '$', 'i')
+          word.match? desired_keyword_as_regexp
+        end
+      end
+    end
+  end
+end

app/models/oclc/lc_call_slips/record.rb

@@ -14,7 +14,7 @@ def generally_relevant?
       end
 
       def relevant_to_selector?(selector:)
-        location_relevant_to_selector?(selector:) && (call_number_in_range_for_selector?(selector:) || subject_relevant_to_selector?(selector:))
+        location_relevant_to_selector?(selector:) && (call_number_in_range_for_selector?(selector:) || subject_relevant_to_selector?(selector:) || keywords_relevant_to_selector?(selector:))
       end
 
       def location_relevant_to_selector?(selector:)
@@ -41,6 +41,12 @@ def class_relevant_to_selector?(selector:)
         false
       end
 
+      def keywords_relevant_to_selector?(selector:)
+        keywords = selector.keywords
+        return false if keywords.blank?
+        record.fields.any? { |field| KeywordField.new(field:, keywords:).match? }
+      end
+
       def subject_relevant_to_selector?(selector:)
         return false unless selector.subjects
 

app/models/oclc/lc_call_slips/selector.rb

@@ -24,6 +24,10 @@ def classes
         call_number_ranges.pluck(:class).uniq
       end
 
+      def keywords
+        @keywords ||= selector_config[selector_key][:keywords]
+      end
+
       def subjects
         selector_config[selector_key][:subjects]
       end

spec/models/oclc/lc_call_slips/keyword_field_spec.rb

@@ -0,0 +1,64 @@
+# frozen_string_literal: true
+require 'rails_helper'
+
+RSpec.describe Oclc::LcCallSlips::KeywordField do
+  shared_examples 'a match' do
+    it 'match? returns true' do
+      expect(described_class.new(field:, keywords:).match?).to eq(true)
+    end
+  end
+  shared_examples 'not a match' do
+    it 'match? returns false' do
+      expect(described_class.new(field:, keywords:).match?).to eq(false)
+    end
+  end
+  context 'when field is a control field' do
+    let(:field) { MARC::ControlField.new('001', 'SCSB-10482146') }
+    let(:keywords) { ['cantaloup*'] }
+    it_behaves_like 'not a match'
+  end
+
+  context 'when field is a 245' do
+    let(:field) do
+      MARC::DataField.new('245', '0', '0',
+            MARC::Subfield.new('a', 'Cantaloups'))
+    end
+    let(:keywords) { ['cantaloup*'] }
+    it_behaves_like 'a match'
+  end
+
+  context 'when field is a 246' do
+    let(:field) do
+      MARC::DataField.new('246', '0', '0',
+            MARC::Subfield.new('a', 'Cantaloupe culture'))
+    end
+    let(:keywords) { ['cantaloup*'] }
+    it_behaves_like 'a match'
+  end
+
+  context 'when field is a 260' do
+    let(:field) do
+      MARC::DataField.new('260', '0', '0',
+            MARC::Subfield.new('b', 'The Rocky Ford Cantaloupe Seed Breeders\' Association'))
+    end
+    let(:keywords) { ['cantaloup*'] }
+    it_behaves_like 'a match'
+    context 'when keyword includes multiple wildcards' do
+      let(:keywords) { ['*nt*loup*'] }
+      it_behaves_like 'a match'
+    end
+    context 'when the first matching keyword comes late in the array' do
+      let(:keywords) { ['sediment', 'metamorphosis', 'geolog*', 'igneous', 'rock*', 'sandstone'] }
+      it_behaves_like 'a match'
+    end
+  end
+
+  context 'when field is a 300' do
+    let(:field) do
+      MARC::DataField.new('300', '0', '0',
+            MARC::Subfield.new('a', '1 online resource (2 pages), digital, PDF file'))
+    end
+    let(:keywords) { ['online'] }
+    it_behaves_like 'not a match'
+  end
+end

spec/models/oclc/lc_call_slips/record_spec.rb

@@ -407,4 +407,69 @@
       expect(oclc_record.generally_relevant?).to eq(true)
     end
   end
+
+  describe 'keywords' do
+    let(:marc_record) { MARC::Record.new_from_hash('fields' => fields) }
+    let(:fields) do
+      [
+        { '008' => '120627s2024    gerabg  ob    001 0 gre d' },
+        { '245' => { "ind1" => "1",
+                     "ind2" => "0",
+                     'subfields' => [
+                       { 'a' => 'Chinese homestyle : ',
+                         'b' => 'everyday plant-based recipes for takeout, dim sum, noodles, and more / ',
+                         'c' => 'Maggie Zhu.' }
+                     ] } }
+      ]
+    end
+    context 'with a selector with a keyword' do
+      let(:selector_config) do
+        { heijdra: {
+          keywords: ['chinese']
+        } }
+      end
+
+      it 'recognizes that it is relevant' do
+        expect(oclc_record.keywords_relevant_to_selector?(selector:)).to eq(true)
+      end
+    end
+    context 'with a selector without any keywords' do
+      let(:selector_config) do
+        { hatfield: {
+          classes: [{ class: 'PN', low_num: 6755, high_num: 6758 }]
+        } }
+      end
+      it 'does not mark a record as relevant based on its subjects' do
+        expect(oclc_record.keywords_relevant_to_selector?(selector:)).to eq(false)
+        expect(oclc_record.relevant_to_selector?(selector:)).to eq(false)
+      end
+    end
+    context 'with wildcard in the config file keywords' do
+      let(:selector_config) do
+        { heijdra: {
+          keywords: ['chin*']
+        } }
+      end
+
+      it 'matches using the wildcard' do
+        expect(oclc_record.keywords_relevant_to_selector?(selector:)).to eq(true)
+      end
+
+      context 'keyword in record contains configured keyword, but config does not have wildcard at beginning and end' do
+        let(:fields) do
+          [
+            { '008' => '120627s2024    gerabg  ob    001 0 gre d' },
+            { '245' => { "ind1" => "1",
+                         "ind2" => "0",
+                         'subfields' => [
+                           { 'a' => 'Machine learning : ' }
+                         ] } }
+          ]
+        end
+        it 'does not mark a record as relevant based on its subjects' do
+          expect(oclc_record.keywords_relevant_to_selector?(selector:)).to eq(false)
+        end
+      end
+    end
+  end
 end