pennmarc 1.0.24 → 1.0.25

checksums.yaml

@@ -1,7 +1,7 @@
 ---
 SHA256:
-  metadata.gz: 59881d630f244bc6d8b76665df84f4b22297de2d51c44f5272df4226f4e27e75
-  data.tar.gz: 8d4e78c359dea40f482adc5b4deaa3fbb75bbe15a4fce5955ad48de054ce82c4
+  metadata.gz: 8b540ecd84333f097b50a3682732229854b188fcde76921eb0ec184622a690dd
+  data.tar.gz: 9203fadee45b9ad65ec20880a1acbc23356000b748d32d07428ec093590cd55a
 SHA512:
-  metadata.gz: 4a3a43ff1cecadf978a8c50d59b382c2592925116cacd757f368ea0a698e9b370a94e74fd98e1a6a06a3b837ab0dbdce2efece991aaf3da4b53c74396a73ee2e
-  data.tar.gz: 7f5468f9891c9fe15003160fb4d5a6c04cd8291ead5e803f0652368376e1fd67381195a7bdffd7ea9a7aa996c071b75240405dd956c46d879753f6811eca1b0a
+  metadata.gz: 6770815e8f5f828c15451fc71c04dbef6a131203e8c58d721c063b15f218d5814a71b37fd6c4f088a9c7164d5d989ff6933abc820094068921f218c937a87410
+  data.tar.gz: 94eee63c29ae06151816a532a5afb7db1f27490de798651c909dc07fd0f33de1b4a1ee26057b6e3a8abc5d15c54d09709df8942e6e21ad828c1296fe3191725e

data/lib/pennmarc/enriched.rb

@@ -12,6 +12,7 @@ module PennMARC
       PHYS_INVENTORY_TAG = 'hld'
       ELEC_INVENTORY_TAG = 'prt'
       ITEM_TAG = 'itm'
+      RELATED_RECORD_TAG = 'rel'
 
       # Subfields for HLD tags
       # Follow MARC 852 spec: https://www.loc.gov/marc/holdings/hd852.html, but names are translated into Alma parlance

data/lib/pennmarc/helpers/access.rb

@@ -5,14 +5,13 @@ module PennMARC
   class Access < Helper
     ONLINE = 'Online'
     AT_THE_LIBRARY = 'At the library'
+    RESOURCE_LINK_BASE_URL = 'hdl.library.upenn.edu'
 
     class << self
       # Based on enhanced metadata fields added by Alma publishing process or API, determine if the record has
       # electronic access or has physical holdings, and is therefore "Online" or "At the library". If a record is "At
       # the library", but has a link to a finding aid in the 856 field (matching certain criteria), also add 'Online' as
       # an access method.
-      # @todo What if none of these criteria match? Should we include "At the library" by default? Records with no value
-      #       in this field would be lost if the user selects a facet value.
       # @param [MARC::Record] record
       # @return [Array]
       def facet(record)
@@ -24,7 +23,7 @@ module PennMARC
         return values if values.size == 2 # return early if all values are already present
 
         # only check if ONLINE isn't already there
-        values << ONLINE if values.exclude?(ONLINE) && finding_aid_linkage?(record)
+        values << ONLINE if values.exclude?(ONLINE) && resource_link?(record)
         values.uniq
       end
 
@@ -44,20 +43,23 @@ module PennMARC
         field.tag.in? [Enriched::Pub::PHYS_INVENTORY_TAG, Enriched::Api::PHYS_INVENTORY_TAG]
       end
 
-      # Check if a record contains an 856 entry for an online finding aid, meeting these criteria:
+      # Check if a record contains an 856 entry with a Penn Handle server link meeting these criteria:
       # 1. Indicator 1 is 4 (HTTP resource)
       # 2. Indicator 2 is NOT 2 (indicating the linkage is to a "related" thing)
       # 3. The URL specified in subfield u (URI) is a Penn Handle link
+      # 4. The subfield z does NOT include the string 'Finding aid'
       # See: https://www.loc.gov/marc/bibliographic/bd856.html
+      # @note Some electronic records do not have Portfolios in Alma, so we rely upon the Resource Link in the 856 to
+      #       get these records included in the Online category.
       # @param [MARC::Record] record
       # @return [Boolean]
-      def finding_aid_linkage?(record)
+      def resource_link?(record)
         record.fields('856').filter_map do |field|
           next if field.indicator2 == '2' || field.indicator1 != '4'
 
           subz = subfield_values(field, 'z')
           subfield_values(field, 'u').filter_map do |value|
-            return true if subz.include?('Finding aid') && value.include?('hdl.library.upenn.edu')
+            return true if subz.exclude?('Finding aid') && value.include?(RESOURCE_LINK_BASE_URL)
           end
         end
         false

data/lib/pennmarc/helpers/identifier.rb

@@ -159,6 +159,18 @@ module PennMARC
         }.uniq
       end
 
+      # Gets any Host record MMS ID values from an Enriched::Pub::RELATED_RECORD_TAG field added during Alma enrichment.
+      # This aids in our handling of "bound with" records.
+      # @param [MARC::Record] record
+      # @return [Array<String>]
+      def host_record_id(record)
+        record.fields(Enriched::Pub::RELATED_RECORD_TAG).filter_map { |field|
+          next unless subfield_value?(field, 'c', /contains/i)
+
+          subfield_values field, :w
+        }.flatten.uniq
+      end
+
       private
 
       # Determine if subfield 'a' is an OCLC id.

data/lib/pennmarc/helpers/subject.rb

@@ -18,7 +18,7 @@ module PennMARC
       # - 2: MeSH
       # - 4: Source not specified (local?)
       # - 7: Source specified in ǂ2
-      SEARCH_SOURCE_INDICATORS = %w[0 1 2 4 7].freeze
+      VALID_SOURCE_INDICATORS = %w[0 1 2 4 7].freeze
 
       # Tags that serve as sources for Subject facet values
       DISPLAY_TAGS = %w[600 610 611 630 650 651].freeze
@@ -27,7 +27,7 @@ module PennMARC
       LOCAL_TAGS = %w[690 691 697].freeze
 
       # All Subjects for searching. This includes most subfield content from any field contained in {SEARCH_TAGS} or
-      # 69X, including any linked 880 fields. Fields must have an indicator2 value in {SEARCH_SOURCE_INDICATORS}.
+      # 69X, including any linked 880 fields. Fields must have an indicator2 value in {VALID_SOURCE_INDICATORS}.
       # @todo this includes subfields that may not be desired like 1 (uri) and 2 (source code) but this might be OK for
       #       a search (non-display) field?
       # @param [Hash] relator_map
@@ -179,12 +179,18 @@ module PennMARC
         end
       end
 
-      # Is a field intended for display in a general subject field
+      # Is a field intended for display in a general subject field? To be included, the field tag is in either
+      # {DISPLAY_TAGS} or {LOCAL_TAGS}, and has an indicator 2 value that is in {VALID_SOURCE_INDICATORS}. If
+      # indicator 2 is '7' - indicating "source specified", the specified source must be in our approved source code
+      # list.
+      # @see Util.valid_subject_genre_source_code?
       # @param [MARC::DataField] field
       # @return [Boolean] whether a MARC field is intended for display under general "Subjects"
       def subject_general_display_field?(field)
         return false unless field.tag.in?(DISPLAY_TAGS + LOCAL_TAGS) && field.respond_to?(:indicator2)
 
+        return false if field.indicator2.present? && !field.indicator2.in?(VALID_SOURCE_INDICATORS)
+
         return false if field.indicator2 == '7' && !valid_subject_genre_source_code?(field)
 
         true
@@ -278,7 +284,7 @@ module PennMARC
       # @param [MARC::DataField] field
       # @return [Boolean]
       def subject_search_field?(field)
-        return false unless field.respond_to?(:indicator2) && SEARCH_SOURCE_INDICATORS.include?(field.indicator2)
+        return false unless field.respond_to?(:indicator2) && VALID_SOURCE_INDICATORS.include?(field.indicator2)
 
         tag = if field.tag == '880'
                 subfield_values(field, '6').first

data/lib/pennmarc/helpers/title.rb

@@ -3,33 +3,36 @@
 module PennMARC
   # This helper contains logic for parsing out Title and Title-related fields.
   class Title < Helper
-    class << self
-      # 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],
-        note: %w[505]
-      }.freeze
+    # 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],
+      note: %w[505]
+    }.freeze
+
+    # This text is used in Alma to indicate a Bib record is a "Host" record for other bibs (bound-withs)
+    HOST_BIB_TITLE = 'Host bibliographic record for boundwith'
 
+    class << self
       # 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
@@ -217,6 +220,16 @@ module PennMARC
         }.uniq
       end
 
+      # Determine if the record is a "Host" bibliographic record for other bib records ("bound-withs")
+      # @param [MARC::Record] record
+      # @return [Boolean]
+      def host_bib_record?(record)
+        record.fields('245').any? do |f|
+          title = join_subfields(f, &subfield_in?(%w[a]))
+          title.include?(HOST_BIB_TITLE)
+        end
+      end
+
       private
 
       # Create prefix/filing hash for representing a title value with filing characters removed, with special

data/lib/pennmarc/version.rb

@@ -1,5 +1,5 @@
 # frozen_string_literal: true
 
 module PennMARC
-  VERSION = '1.0.24'
+  VERSION = '1.0.25'
 end

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

@@ -69,6 +69,17 @@ describe 'PennMARC::Access' do
           { z: 'Finding aid', u: 'http://hdl.library.upenn.edu/1017/d/pacscl/UPENN_RBML_MsColl200' }
         end
 
+        it 'does not include online access' do
+          expect(helper.facet(record)).to contain_exactly(PennMARC::Access::AT_THE_LIBRARY)
+        end
+      end
+
+      context 'with an 856 describing a resource link' do
+        let(:indicators) { { indicator1: '4', indicator2: '1' } }
+        let(:location_and_access_subfields) do
+          { z: 'Connect to resource', u: 'http://hdl.library.upenn.edu/1234' }
+        end
+
         it 'includes online access' do
           expect(helper.facet(record)).to contain_exactly(PennMARC::Access::ONLINE, PennMARC::Access::AT_THE_LIBRARY)
         end

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

@@ -161,4 +161,18 @@ describe 'PennMARC::Identifier' do
                                                          '10.1038/sdata.2016.18', '10.18574/9781479842865')
     end
   end
+
+  describe '.host_record_id' do
+    let(:record) do
+      marc_record fields: [
+        marc_field(tag: PennMARC::Enriched::Pub::RELATED_RECORD_TAG, subfields: { w: '123456789', c: 'Contains',
+                                                                                  a: 'Title' }),
+        marc_field(tag: PennMARC::Enriched::Pub::RELATED_RECORD_TAG, subfields: { w: '666666666', c: 'Contained In' })
+      ]
+    end
+
+    it 'returns only the desired host record MMS ID values' do
+      expect(helper.host_record_id(record)).to contain_exactly '123456789'
+    end
+  end
 end

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

@@ -288,6 +288,22 @@ describe 'PennMARC::Subject' do
                                    'Early works to 1950.')
       end
     end
+
+    context 'with a mix of allowed and disallowed heading sources' do
+      let(:fields) do
+        [marc_field(tag: '650', indicator2: '7', subfields: {
+                      '2': 'fast',
+                      a: 'Philosophy in motion pictures.',
+                      '0': 'http://id.loc.gov/authorities/subjects/sh92003501'
+                    }),
+         marc_field(tag: '650', indicator2: '7', subfields: { '2': 'gnd', a: 'Filmästhetik.' }),
+         marc_field(tag: '650', indicator2: '6', subfields: { a: 'Cinéma et arts.' })]
+      end
+
+      it 'includes only permitted headings from approved ontologies' do
+        expect(values).to contain_exactly 'Philosophy in motion pictures.'
+      end
+    end
   end
 
   describe '.childrens_show' do

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

@@ -249,4 +249,24 @@ describe 'PennMARC::Title' do
       expect(values).not_to include 'Linkage', '247'
     end
   end
+
+  describe '.host_bib_record?' do
+    let(:record) { marc_record fields: [marc_field(tag: '245', subfields: subfields)] }
+
+    context 'with a host record' do
+      let(:subfields) { { a: "#{PennMARC::Title::HOST_BIB_TITLE} for 123456789" } }
+
+      it 'returns true' do
+        expect(helper.host_bib_record?(record)).to be true
+      end
+    end
+
+    context 'with a non-host record' do
+      let(:subfields) { { a: 'Regular record' } }
+
+      it 'returns false' do
+        expect(helper.host_bib_record?(record)).to be false
+      end
+    end
+  end
 end

metadata

@@ -1,7 +1,7 @@
 --- !ruby/object:Gem::Specification
 name: pennmarc
 version: !ruby/object:Gem::Version
-  version: 1.0.24
+  version: 1.0.25
 platform: ruby
 authors:
 - Mike Kanning
@@ -10,7 +10,7 @@ authors:
 autorequire:
 bindir: bin
 cert_chain: []
-date: 2024-06-18 00:00:00.000000000 Z
+date: 2024-06-26 00:00:00.000000000 Z
 dependencies:
 - !ruby/object:Gem::Dependency
   name: activesupport