pennmarc 1.0.6 → 1.0.7

checksums.yaml

@@ -1,7 +1,7 @@
 ---
 SHA256:
-  metadata.gz: cb8a665310aaf17d406fcfbfe562894a71f17d35b2d63173ca587a9c0e343b54
-  data.tar.gz: 671549b45eb522db5f4350aa0743bcd16585398c6c07f71e2d2a772367a0f33c
+  metadata.gz: 11dd403cd5a0f964db340be6258d2cc510c99ccc0fc6d43a65b236853cb5b87c
+  data.tar.gz: 85eed79f3e637121cce8be8b0f0509d26e3fa9b6283b7f5bc48b0b321259e3f3
 SHA512:
-  metadata.gz: 9a1374b40186e61a56e33924a6f76561877af209e3b8486b185599766e5cc7bc1b45362cf4f3f9e18fc5cf2cb4328b46f8b0f83b0e2c374a1f31164bce3b8483
-  data.tar.gz: 7a098756899b310a694dbcc1409516de62ff9897b624ac4783b793de145ac61c497bc183409343269e5e33b7b21a07728fa74148fe6af2ad8aaef8149acd71c5
+  metadata.gz: 90a59eaf167d985b596663d6b8d1eab0c3263724da550babfb28fa253df326b07b72bbb5aac1cf1d4150efd2eaecce84686b2de9e99801ff178bbfabc982b765
+  data.tar.gz: 5e0a4fdc1518af2b3a52bc137f76a30430d025f58f5cd46b9d0347a7a6a08f901e72e45917cc9a52d0352a6dbb0213725fd5a6e22650cfedb00d534fb72b44f2

data/README.md

@@ -71,7 +71,7 @@ rspec
 
 ## Publishing the Gem
 
-1. Update the version in `pennmarc.gemspec`
+1. Update the `VERSION` constant in `lib/pennmarc/version.rb`.
 2. Run `gem build pennmarc.gemspec` with the latest code
 3. Run `gem push pennmarc-{version number here}.gem`(e.g. `gem push pennmarc-1.0.0.gem`) to push to RubyGems. You will need access and MFA setup with RubyGems.
 

data/lib/pennmarc/helpers/language.rb

@@ -26,9 +26,9 @@ module PennMARC
       # when no linguistic content is found.
       #
       # @note In franklin, we extracted the language code from the 008 control field. After engaging cataloging unit
-      #   representatives, we decided to extract these values from the 041 field: Includes records for multilingual
-      #   items, items that involve translation, and items where the medium of communication is a sign language.
-      #   https://www.loc.gov/marc/bibliographic/bd041.html
+      #   representatives, we decided to also extract these values from the 041 field: Includes records for
+      #   multilingual items, items that involve translation, and items where the medium of communication is a sign
+      #   language. https://www.loc.gov/marc/bibliographic/bd041.html
       #
       # @param [MARC::Record] record
       # @param [Hash] iso_639_2_mapping iso-639-2 spec hash for language code translation

data/lib/pennmarc/helpers/link.rb

@@ -10,9 +10,52 @@ module PennMARC
       # @return [Object]
       def offsite(record); end
 
-      def full_text(record:); end
+      # Full text links from MARC 856 fields.
+      # @param [MARC::Record] record
+      # @return [Array] array of hashes
+      def full_text(record:)
+        indicator2_options = %w[0 1]
+        links_from_record(record, indicator2_options)
+      end
+
+      # Web text links from MARC 856 fields.
+      # @param [MARC::Record] record
+      # @return [Array] array of hashes
+      def web(record:)
+        indicator2_options = ['2', ' ', '']
+        links_from_record(record, indicator2_options)
+      end
+
+      private
+
+      # Extract subfield 3 and z/y depending on the presence of either. Extract link url and assemble array
+      # with text and link.
+      # @param [MARC::Field] field
+      # @return [Array]
+      def link_text_and_url(field)
+        subfield3 = subfield_values(field, 3)
+        subfield_zy = field.find_all(&subfield_in?(%w[z y])).map(&:value)
+        link_text = [subfield3, subfield_zy.first].compact.join(' ')
+        link_url = subfield_values(field, 'u')&.first || ''
+        [link_text, link_url.sub(' target=_blank', '')]
+      end
+
+      # Assemble array of link text, link URL values from 856 fields. Ensure indicator1 (access method)
+      # is always 4 (HTTP) and indicator2 (relationship) can be specified by caller method.
+      # @param [MARC::Record] record
+      # @param [Array] indicator2_options
+      # @return [Array]
+      def links_from_record(record, indicator2_options)
+        record.fields('856').filter_map do |field|
+          next unless field.indicator1 == '4' && indicator2_options.include?(field.indicator2)
 
-      def web(record:); end
+          link_text, link_url = link_text_and_url(field)
+          {
+            link_text: link_text.present? ? link_text : link_url,
+            link_url: link_url
+          }
+        end
+      end
     end
   end
 end

data/lib/pennmarc/helpers/title.rb

@@ -4,46 +4,81 @@ module PennMARC
   # This helper contains logic for parsing out Title and Title-related fields.
   class Title < Helper
     class << self
-      # these will be used when completing the *search_aux methods
+      # We use these fields when retrieving auxiliary titles in the *search_aux methods:
+      # {https://www.loc.gov/marc/bibliographic/bd130.html 130},
+      # {https://www.loc.gov/marc/bibliographic/bd210.html 210},
+      # {https://www.loc.gov/marc/bibliographic/bd245.html 245},
+      # {https://www.loc.gov/marc/bibliographic/bd246.html 246},
+      # {https://www.loc.gov/marc/bibliographic/bd247.html 247},
+      # {https://www.loc.gov/marc/bibliographic/bd440.html 440},
+      # {https://www.loc.gov/marc/bibliographic/bd490.html 490},
+      # {https://www.loc.gov/marc/bibliographic/bd730.html 730},
+      # {https://www.loc.gov/marc/bibliographic/bd740.html 740},
+      # {https://www.loc.gov/marc/bibliographic/bd830.html 830},
+      # {https://www.loc.gov/marc/bibliographic/bd773.html 773},
+      # {https://www.loc.gov/marc/bibliographic/bd774.html 774},
+      # {https://www.loc.gov/marc/bibliographic/bd780.html 780},
+      # {https://www.loc.gov/marc/bibliographic/bd785.html 785},
+      # {https://www.loc.gov/marc/bibliographic/bd700.html 700},
+      # {https://www.loc.gov/marc/bibliographic/bd710.html 710},
+      # {https://www.loc.gov/marc/bibliographic/bd711.html 711},
+      # {https://www.loc.gov/marc/bibliographic/bd505.html 505}
       AUX_TITLE_TAGS = {
         main: %w[130 210 240 245 246 247 440 490 730 740 830],
         related: %w[773 774 780 785],
-        entity: %w[700 710 711]
+        entity: %w[700 710 711],
+        note: %w[505]
       }.freeze
 
-      # Main Title Search field. Takes from 245 and linked 880.
+      # Main Title Search field. Takes from {https://www.loc.gov/marc/bibliographic/bd245.html 245} and linked 880.
       # @note Ported from get_title_1_search_values.
       # @param [MARC::Record] record
       # @return [Array<String>] array of title values for search
       def search(record)
-        titles = record.fields('245').filter_map do |field|
-          join_subfields(field, &subfield_not_in?(%w[c 6 8 h]))
-        end
-        titles + record.fields('880').filter_map do |field|
-          next unless subfield_value?(field, '6', /245/)
+        record.fields(%w[245 880]).filter_map do |field|
+          next if field.tag == '880' && subfield_value_not_in?(field, '6', %w[245])
 
           join_subfields(field, &subfield_not_in?(%w[c 6 8 h]))
         end
       end
 
-      # Auxiliary Title Search field. Takes from many fields that contain title-like information.
-      # @note Ported from get_title_2_search_values.
-      # @todo port this, it is way complicated but essential for relevance
+      # Auxiliary Title Search field. Takes from many fields defined in {AUX_TITLE_TAGS} that contain title-like
+      # information.
       # @param [MARC::Record] record
-      # @return [Array<String>] array of title values for search
-      def search_aux(record); end
+      # @return [Array<String>] array of auxiliary title values for search
+      def search_aux(record)
+        search_aux_values(record: record, title_type: :main, &subfield_not_in?(%w[c 6 8])) +
+          search_aux_values(record: record, title_type: :related, &subfield_not_in?(%w[s t])) +
+          search_aux_values(record: record, title_type: :entity, &subfield_in?(%w[t])) +
+          search_aux_values(record: record, title_type: :note, &subfield_in?(%w[t]))
+      end
 
-      # Journal Title Search field.
-      # @todo port this, it is way complicated but essential for relevance
+      # Journal Title Search field. Takes from {https://www.loc.gov/marc/bibliographic/bd245.html 245} and linked 880.
+      # We do not return any values if the {https://www.loc.gov/marc/bibliographic/bdleader.html MARC leader}
+      # indicates that the record is not a serial.
       # @param [MARC::Record] record
       # @return [Array<String>] journal title information for search
-      def journal_search(record); end
+      def journal_search(record)
+        return [] if not_a_serial?(record)
+
+        record.fields(%w[245 880]).filter_map do |field|
+          next if field.tag == '880' && subfield_value_not_in?(field, '6', %w[245])
+
+          join_subfields(field, &subfield_not_in?(%w[c 6 8 h]))
+        end
+      end
 
-      # Auxiliary Journal Title Search field.
-      # @todo port this, it is way complicated but essential for relevance
+      # Auxiliary Journal Title Search field. Takes from many fields defined in {AUX_TITLE_TAGS} that contain title-like
+      # information. Does not return any titles if the {https://www.loc.gov/marc/bibliographic/bdleader.html MARC leader}
+      # indicates that the record is not a serial.
       # @param [MARC::Record] record
-      # @return [Array<String>] journal title information for search
-      def journal_search_aux(record); end
+      # @return [Array<String>] auxiliary journal title information for search
+      def journal_search_aux(record)
+        search_aux_values(record: record, title_type: :main, journal: true, &subfield_not_in?(%w[c 6 8])) +
+          search_aux_values(record: record, title_type: :related, journal: true, &subfield_not_in?(%w[s t])) +
+          search_aux_values(record: record, title_type: :entity, journal: true, &subfield_in?(%w[t])) +
+          search_aux_values(record: record, title_type: :note, journal: true, &subfield_in?(%w[t]))
+      end
 
       # Single-valued Title, for use in headings. Takes the first {https://www.oclc.org/bibformats/en/2xx/245.html 245}
       # value. Special consideration for
@@ -192,6 +227,43 @@ module PennMARC
           { prefix: '', filing: title.strip }
         end
       end
+
+      # Evaluate {https://www.loc.gov/marc/bibliographic/bdleader.html MARC leader} to determine if record is a serial.
+      # @param [MARC::Record] record
+      # @return [Boolean]
+      def not_a_serial?(record)
+        !record.leader[6..7].ends_with?('s')
+      end
+
+      # @param [MARC::DataField] field
+      # @param [String] value
+      # @return [Boolean]
+      def indicators_are_not_value?(field, value)
+        field.indicator1 != value && field.indicator2 != value
+      end
+
+      # Retrieve auxiliary title values. Returns no values if a journal is expected but the
+      # {https://www.loc.gov/marc/bibliographic/bdleader.html MARC leader} indicates that the record is not a serial.
+      # We take special consideration for the {https://www.loc.gov/marc/bibliographic/bd505.html 505 field}, extracting
+      # values only when indicator1 and indicator2 are both '0'.
+      # @param [MARC::Record] record
+      # @param [Symbol] title_type
+      # @param [Boolean] journal
+      # @bloc [Proc] join_selector
+      # @return [Array<String>]
+      def search_aux_values(record:, title_type:, journal: false, &join_selector)
+        return [] if journal && not_a_serial?(record)
+
+        tags = AUX_TITLE_TAGS[title_type] + ['880']
+
+        record.fields(tags).filter_map do |field|
+          next if field.tag == '505' && indicators_are_not_value?(field, '0')
+
+          next if field.tag == '880' && subfield_value_not_in?(field, '6', tags)
+
+          join_subfields(field, &join_selector)
+        end
+      end
     end
   end
 end

data/lib/pennmarc/version.rb

@@ -1,5 +1,5 @@
 # frozen_string_literal: true
 
 module PennMARC
-  VERSION = '1.0.6'
+  VERSION = '1.0.7'
 end

data/spec/lib/pennmarc/helpers/link_spec.rb

@@ -0,0 +1,37 @@
+# frozen_string_literal: true
+
+describe 'PennMARC::Link' do
+  include MarcSpecHelpers
+
+  let(:helper) { PennMARC::Link }
+
+  describe '.full_text' do
+    let(:record) do
+      marc_record fields: [marc_field(tag: '856', subfields: { '3': 'Materials specified',
+                                                               z: 'Public note',
+                                                               y: 'Link text',
+                                                               u: 'https://www.test-uri.com/' },
+                                      indicator1: '4', indicator2: '0')]
+    end
+
+    it 'returns full text link text and url' do
+      expect(helper.full_text(record: record)).to contain_exactly({ link_text: 'Materials specified Public note',
+                                                                    link_url: 'https://www.test-uri.com/' })
+    end
+  end
+
+  describe '.web' do
+    let(:record) do
+      marc_record fields: [marc_field(tag: '856', subfields: { '3': 'Materials specified',
+                                                               z: 'Public note',
+                                                               y: 'Link text',
+                                                               u: 'https://www.test-uri.com/' },
+                                      indicator1: '4', indicator2: '')]
+    end
+
+    it 'returns web link text and url' do
+      expect(helper.web(record: record)).to contain_exactly({ link_text: 'Materials specified Public note',
+                                                              link_url: 'https://www.test-uri.com/' })
+    end
+  end
+end

data/spec/lib/pennmarc/helpers/title_spec.rb

@@ -21,7 +21,86 @@ describe 'PennMARC::Title' do
   end
 
   describe '.search_aux' do
-    it 'returns search aux values', pending: 'Not implemented yet'
+    let(:leader) { 'ZZZZZnaaZa22ZZZZZzZZ4500' }
+    let(:record) do
+      marc_record fields: [
+        marc_field(tag: '130', subfields: { a: 'Uniform Title', c: '130 not included' }),
+        marc_field(tag: '880', subfields: { '6': '130', a: 'Alternative Uniform Title' }),
+        marc_field(tag: '773', subfields: { a: 'Host Uniform Title', s: '773 not included' }),
+        marc_field(tag: '700', subfields: { t: 'Personal Entry Title', s: '700 not included' }),
+        marc_field(tag: '505', subfields: { t: 'Invalid Formatted Contents Note Title' }, indicator1: 'invalid'),
+        marc_field(tag: '505', subfields: { t: 'Formatted Contents Note Title', s: '505 not included' },
+                   indicator1: '0', indicator2: '0')
+      ], leader: leader
+    end
+
+    it 'returns auxiliary titles' do
+      expect(helper.search_aux(record)).to contain_exactly('Uniform Title', 'Host Uniform Title',
+                                                           'Alternative Uniform Title', 'Personal Entry Title',
+                                                           'Formatted Contents Note Title')
+    end
+
+    context 'when the leader indicates the record is a serial' do
+      let(:leader) { 'ZZZZZnasZa22ZZZZZzZZ4500' }
+
+      it 'returns auxiliary titles' do
+        expect(helper.search_aux(record)).to contain_exactly('Uniform Title', 'Host Uniform Title',
+                                                             'Alternative Uniform Title', 'Personal Entry Title',
+                                                             'Formatted Contents Note Title')
+      end
+    end
+  end
+
+  describe '.journal_search' do
+    let(:leader) { 'ZZZZZnasZa22ZZZZZzZZ4500' }
+    let(:record) do
+      marc_record fields: [
+        marc_field(tag: '245', subfields: { a: 'Some Journal Title' }),
+        marc_field(tag: '880', subfields: { a: 'Alternative Script', '6': '245' }),
+        marc_field(tag: '880', subfields: { a: 'Unrelated 880', '6': 'invalid' })
+      ], leader: leader
+    end
+
+    it 'returns journal search titles' do
+      expect(helper.journal_search(record)).to contain_exactly('Some Journal Title', 'Alternative Script')
+    end
+
+    context 'when the record is not a serial' do
+      let(:leader) { 'ZZZZZnaaZa22ZZZZZzZZ4500' }
+
+      it 'returns an empty array' do
+        expect(helper.journal_search_aux(record)).to be_empty
+      end
+    end
+  end
+
+  describe '.journal_search_aux' do
+    let(:leader) { 'ZZZZZnasZa22ZZZZZzZZ4500' }
+    let(:record) do
+      marc_record fields: [
+        marc_field(tag: '130', subfields: { a: 'Uniform Title', c: '130 not included' }),
+        marc_field(tag: '880', subfields: { '6': '130', a: 'Alternative Uniform Title' }),
+        marc_field(tag: '773', subfields: { a: 'Host Uniform Title', s: '773 not included' }),
+        marc_field(tag: '700', subfields: { t: 'Personal Entry Title', s: '700 not included' }),
+        marc_field(tag: '505', subfields: { t: 'Invalid Formatted Contents Note Title' }, indicator1: 'invalid'),
+        marc_field(tag: '505', subfields: { t: 'Formatted Contents Note Title', s: '505 not included' },
+                   indicator1: '0', indicator2: '0')
+      ], leader: leader
+    end
+
+    it 'returns auxiliary journal search titles' do
+      expect(helper.journal_search_aux(record)).to contain_exactly('Uniform Title', 'Alternative Uniform Title',
+                                                                   'Host Uniform Title', 'Personal Entry Title',
+                                                                   'Formatted Contents Note Title')
+    end
+
+    context 'when the record is not a serial' do
+      let(:leader) { 'ZZZZZnaaZa22ZZZZZzZZ4500' }
+
+      it 'returns an empty array' do
+        expect(helper.journal_search_aux(record)).to be_empty
+      end
+    end
   end
 
   describe '.show' do

metadata

@@ -1,7 +1,7 @@
 --- !ruby/object:Gem::Specification
 name: pennmarc
 version: !ruby/object:Gem::Version
-  version: 1.0.6
+  version: 1.0.7
 platform: ruby
 authors:
 - Mike Kanning
@@ -10,7 +10,7 @@ authors:
 autorequire:
 bindir: bin
 cert_chain: []
-date: 2023-11-09 00:00:00.000000000 Z
+date: 2023-12-06 00:00:00.000000000 Z
 dependencies:
 - !ruby/object:Gem::Dependency
   name: activesupport
@@ -128,6 +128,7 @@ files:
 - spec/lib/pennmarc/helpers/genre_spec.rb
 - spec/lib/pennmarc/helpers/identifer_spec.rb
 - spec/lib/pennmarc/helpers/language_spec.rb
+- spec/lib/pennmarc/helpers/link_spec.rb
 - spec/lib/pennmarc/helpers/location_spec.rb
 - spec/lib/pennmarc/helpers/note_spec.rb
 - spec/lib/pennmarc/helpers/production_spec.rb