pennmarc 1.0.26 → 1.0.30

checksums.yaml

@@ -1,7 +1,7 @@
 ---
 SHA256:
-  metadata.gz: 8b4d6a53347f4532faa566debcc59bde1fb7043cfbd4530d0a06489a36db8599
-  data.tar.gz: 8a92c36c80406541ac454025e6c2adc53af80dab20af5af16b8995dc5bacfe9d
+  metadata.gz: '07810ba2a6de07713ce40a1cfbe7fc33d5497eec834011a0098ce2cbd74bc5ff'
+  data.tar.gz: 45c3f0002e46be338a63be2109153f7603d9ebf8b1814d825b4c889c7f7237b4
 SHA512:
-  metadata.gz: 8b1f0be0caba6a00adc8ccc5516556b2de250d93861157c49a60e17f52810b6eed7942efe8b51843d7428bb329c6dcad58a0887c17e9132004fe3d7d29729682
-  data.tar.gz: b06dad73899b13254391b74832d3b99745d82bc9f871801d7a617e2416f9d70039d45a9122a292377628b638c4044e692432688dbce48497cb6394fa48ba0d6e
+  metadata.gz: 9a214837d1f061611a13ea7d09b13ae41d7f6f7df8f08d03da3d98d8969849aacb0edfd45a3fea231acb1e53f8c06c4e4793db49cf5274063d6c59d7b7043212
+  data.tar.gz: 1a6dd99e562ef38b301e49ac17b6ec211eb0af1e7bdccc6a0d98c15683163f9c4168e51b8560296282a42d6dacdbfbb5d8856628af7000614e48961cbcb6b19c

data/Gemfile

@@ -5,6 +5,7 @@ source 'https://rubygems.org'
 gem 'activesupport', '~> 7'
 gem 'library_stdnums', '~> 1.6'
 gem 'marc', '~> 1.2'
+gem 'multi_string_replace', '~> 2.0'
 gem 'nokogiri', '~> 1.15'
 gem 'rake', '~> 13.0'
 gem 'upennlib-rubocop', require: false

data/Gemfile.lock

@@ -19,6 +19,7 @@ GEM
       scrub_rb (>= 1.0.1, < 2)
       unf
     minitest (5.18.0)
+    multi_string_replace (2.0.2)
     nokogiri (1.15.2-arm64-darwin)
       racc (~> 1.4)
     nokogiri (1.15.2-x64-mingw-ucrt)
@@ -113,6 +114,7 @@ DEPENDENCIES
   activesupport (~> 7)
   library_stdnums (~> 1.6)
   marc (~> 1.2)
+  multi_string_replace (~> 2.0)
   nokogiri (~> 1.15)
   rake (~> 13.0)
   rspec (~> 3.12)

data/README.md

@@ -7,13 +7,28 @@ the "Nouveau Franklin" project aka [discovery_app](https://gitlab.library.upenn.
 When included in a project, it should be utilized like this:
 
 ```ruby
-parser = PennMARC::Parser.new # eventually we will pass in some mappings...
+parser = PennMARC::Parser.new
 puts parser.title_show(marc_record) # Title intended for display
 ```
 
 All methods will require a `MARC::Record` object. For more about these, see the 
 [ruby-marc](https://github.com/ruby-marc/ruby-marc) gem documentation
 
+## Term Overriding
+
+This gem provides configuration as well as a method for overriding and removing terms that are undesirable. In your app,
+you can remove or replace the configured terms like so:
+
+```ruby
+improved_values = PennMARC::HeadingControl.term_override(values)
+```
+
+This will remove any elements of the `values` array that include any terms defined in `mappers/headings_remove.yml` and
+replace any terms defined in the `headings_override.yml` file.
+
+By default, terms are replaced for `Subject#*show` and `Subject#facet` methods. You can bypass the default overriding on
+on these methods by passing `override: false`.
+
 ## Development
 
 ### Requirements

data/lib/pennmarc/enriched.rb

@@ -12,7 +12,7 @@ module PennMARC
       PHYS_INVENTORY_TAG = 'hld'
       ELEC_INVENTORY_TAG = 'prt'
       ITEM_TAG = 'itm'
-      RELATED_RECORD_TAG = 'rel'
+      RELATED_RECORD_TAGS = %w[REL rel].freeze
 
       # 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/heading_control.rb

@@ -1,11 +1,29 @@
 # frozen_string_literal: true
 
+require 'multi_string_replace'
+
 module PennMARC
-  # Shared values for controlling inclusion of subject or genre headings
-  module HeadingControl
-    # These codes are expected to be found in sf2 when the indicator2 value is 7, indicating "source specified". There
-    # are some sources whose headings we don't want to display.
+  # Shared tools and values for controlling handling of subject or genre headings
+  class HeadingControl
+    # These codes are expected to be found in sf2 of a subject/genre field when the indicator2 value is 7, indicating
+    # "source specified". There are some sources whose headings we don't want to display.
     ALLOWED_SOURCE_CODES = %w[aat cct fast ftamc gmgpc gsafd homoit jlabsh lcgft lcsh lcstt lctgm
                               local/osu mesh ndlsh nli nlksh rbbin rbgenr rbmscv rbpap rbpri rbprov rbpub rbtyp].freeze
+
+    class << self
+      # Replace or remove any terms in provided values pursuant to the configuration in remove and override mappers.
+      # Used to remove or replace offensive or otherwise undesirable subject headings.
+      # @param values [Array]
+      # @return [Array] values with terms removed/replaced
+      def term_override(values)
+        values.filter_map do |value|
+          # Remove values if they contain a remove term
+          next nil if value.match?(/#{Mappers.headings_to_remove&.join('|')}/i)
+
+          # Replace values using multi_string_replace gem
+          MultiStringReplace.replace value, Mappers.heading_overrides
+        end
+      end
+    end
   end
 end

data/lib/pennmarc/helpers/identifier.rb

@@ -164,7 +164,7 @@ module PennMARC
       # @param [MARC::Record] record
       # @return [Array<String>]
       def host_record_id(record)
-        record.fields(Enriched::Pub::RELATED_RECORD_TAG).filter_map { |field|
+        record.fields(Enriched::Pub::RELATED_RECORD_TAGS).filter_map { |field|
           next unless subfield_value?(field, 'c', /contains/i)
 
           subfield_values field, :w

data/lib/pennmarc/helpers/subject.rb

@@ -62,67 +62,76 @@ module PennMARC
       #
       # @note this is ported mostly form MG's new-style Subject parsing
       # @param [MARC::Record] record
+      # @param [Boolean] override to remove undesirable terms or not
       # @return [Array<String>] array of all subject values for faceting
-      def facet(record)
-        subject_fields(record, type: :facet).filter_map { |field|
+      def facet(record, override: true)
+        values = subject_fields(record, type: :facet).filter_map { |field|
           term_hash = build_subject_hash(field)
           next if term_hash.blank? || term_hash[:count]&.zero?
 
           format_term type: :facet, term: term_hash
         }.uniq
+        override ? HeadingControl.term_override(values) : values
       end
 
       # All Subjects for display. This includes all {DISPLAY_TAGS} and {LOCAL_TAGS}. For tags that specify a source,
       # only those with an allowed source code (see ALLOWED_SOURCE_CODES) are included.
       #
       # @param [MARC::Record] record
+      # @param [Boolean] override to remove undesirable terms or not
       # @return [Array] array of all subject values for display
-      def show(record)
-        subject_fields(record, type: :all).filter_map { |field|
+      def show(record, override: true)
+        values = subject_fields(record, type: :all).filter_map { |field|
           term_hash = build_subject_hash(field)
           next if term_hash.blank? || term_hash[:count]&.zero?
 
           format_term type: :display, term: term_hash
         }.uniq
+        override ? HeadingControl.term_override(values) : values
       end
 
       # Get Subjects from "Children" ontology
       #
       # @param [MARC::Record] record
+      # @param [Boolean] override to remove undesirable terms or not
       # @return [Array] array of children's subject values for display
-      def childrens_show(record)
-        subject_fields(record, type: :display, options: { tags: DISPLAY_TAGS, indicator2: '1' })
-          .filter_map { |field|
-            term_hash = build_subject_hash(field)
-            next if term_hash.blank? || term_hash[:count]&.zero?
-
-            format_term type: :display, term: term_hash
-          }.uniq
+      def childrens_show(record, override: true)
+        values = subject_fields(record, type: :display, options: { tags: DISPLAY_TAGS, indicator2: '1' })
+                 .filter_map { |field|
+          term_hash = build_subject_hash(field)
+          next if term_hash.blank? || term_hash[:count]&.zero?
+
+          format_term type: :display, term: term_hash
+        }.uniq
+        override ? HeadingControl.term_override(values) : values
       end
 
       # Get Subjects from "MeSH" ontology
       #
       # @param [MARC::Record] record
+      # @param [Boolean] override to remove undesirable terms or not
       # @return [Array] array of MeSH subject values for display
-      def medical_show(record)
-        subject_fields(record, type: :display, options: { tags: DISPLAY_TAGS, indicator2: '2' })
-          .filter_map { |field|
-            term_hash = build_subject_hash(field)
-            next if term_hash.blank? || term_hash[:count]&.zero?
-
-            format_term type: :display, term: term_hash
-          }.uniq
+      def medical_show(record, override: true)
+        values = subject_fields(record, type: :display, options: { tags: DISPLAY_TAGS, indicator2: '2' })
+                 .filter_map { |field|
+          term_hash = build_subject_hash(field)
+          next if term_hash.blank? || term_hash[:count]&.zero?
+
+          format_term type: :display, term: term_hash
+        }.uniq
+        override ? HeadingControl.term_override(values) : values
       end
 
       # Get Subject values from {DISPLAY_TAGS} where indicator2 is 4 and {LOCAL_TAGS}. Do not include any values where
       # sf2 includes "penncoi" (Community of Interest).
       #
       # @param [MARC::Record] record
+      # @param [Boolean] override to remove undesirable terms
       # @return [Array] array of local subject values for display
-      def local_show(record)
+      def local_show(record, override: true)
         local_fields = subject_fields(record, type: :display, options: { tags: DISPLAY_TAGS, indicator2: '4' }) +
                        subject_fields(record, type: :local)
-        local_fields.filter_map { |field|
+        values = local_fields.filter_map { |field|
           next if subfield_value?(field, '2', /penncoi/)
 
           term_hash = build_subject_hash(field)
@@ -130,6 +139,7 @@ module PennMARC
 
           format_term type: :display, term: term_hash
         }.uniq
+        override ? HeadingControl.term_override(values) : values
       end
 
       private

data/lib/pennmarc/mappers.rb

@@ -4,46 +4,58 @@ module PennMARC
   # reusable static mappers
   class Mappers
     class << self
-      # @return [Hash]
+      # @return [Hash, nil]
+      def heading_overrides
+        @heading_overrides ||= load_map('headings_override.yml', symbolize_names: false)
+      end
+
+      # @return [Hash, nil]
+      def headings_to_remove
+        @headings_to_remove ||= load_map('headings_remove.yml', symbolize_names: false)
+      end
+
+      # @return [Hash, nil]
       def iso_639_2_language
         @iso_639_2_language ||= load_map('iso639-2-languages.yml')
       end
 
+      # @return [Hash, nil]
       def iso_639_3_language
         @iso_639_3_language ||= load_map('iso639-3-languages.yml')
       end
 
-      # @return [Hash]
+      # @return [Hash, nil]
       def location
         @location ||= load_map('locations.yml')
       end
 
-      # @return [Hash]
+      # @return [Hash, nil]
       def location_overrides
         @location_overrides ||= load_map('location_overrides.yml')
       end
 
-      # @return [Hash]
+      # @return [Hash, nil]
       def relator
         @relator ||= load_map('relator.yml')
       end
 
-      # @return [Hash]
+      # @return [Hash, nil]
       def loc_classification
         @loc_classification ||= load_map('loc_classification.yml')
       end
 
-      # @return [Hash]
+      # @return [Hash, nil]
       def dewey_classification
         @dewey_classification ||= load_map('dewey_classification.yml')
       end
 
-      # @param [String] filename of mapping file in config directory, with file extension
-      # @return [Hash] mapping as hash
-      def load_map(filename)
+      # @param filename [String] name of mapping file in config directory, with file extension
+      # @param symbolize_names [Boolean] whether or not to symbolize keys in returned hash
+      # @return [Hash, nil] mapping as hash
+      def load_map(filename, symbolize_names: true)
         puts { "Loading #{filename}" }
         YAML.safe_load(File.read(File.join(File.expand_path(__dir__), 'mappings', filename)),
-                       symbolize_names: true)
+                       symbolize_names: symbolize_names)
       end
     end
   end

data/lib/pennmarc/mappings/headings_override.yml

@@ -0,0 +1,8 @@
+Aliens: Noncitizens
+"Alien criminals": Noncitizen criminals
+"Alien detention centers": Immigrant detention centers
+"Alien labor": Foreign workers
+"Alien property": Foreign-owned property
+Gypsies: Romanies
+"Illegal Alien Children": Undocumented immigrant children
+"Illegal Aliens": Undocumented immigrants

data/lib/pennmarc/mappings/headings_remove.yml

@@ -0,0 +1,2 @@
+- Jewish Question
+- Yellow Peril

data/lib/pennmarc/version.rb

@@ -1,5 +1,5 @@
 # frozen_string_literal: true
 
 module PennMARC
-  VERSION = '1.0.26'
+  VERSION = '1.0.30'
 end

data/pennmarc.gemspec

@@ -10,7 +10,7 @@ Gem::Specification.new do |s|
   s.summary     = 'Penn Libraries Catalog MARC parsing wisdom for cross-project usage'
   s.description = 'This gem provides methods for parsing a Penn Libraries MARCXML record into string, array and date
                    objects for use in discovery or preservation applications.'
-  s.authors     = ['Mike Kanning', 'Amrey Mathurin', 'Patrick Perkins']
+  s.authors     = ['Mike Kanning', 'Amrey Mathurin', 'Patrick Perkins', 'Katherine Schultz', 'Baowei Wei']
   s.email       = 'mkanning@upenn.edu'
   s.files       = `git ls-files`.split($OUTPUT_RECORD_SEPARATOR)
   s.homepage    = 'https://gitlab.library.upenn.edu/dld/catalog/pennmarc'
@@ -21,6 +21,7 @@ Gem::Specification.new do |s|
   s.add_dependency 'activesupport', '~> 7'
   s.add_dependency 'library_stdnums', '~> 1.6'
   s.add_dependency 'marc', '~> 1.2'
+  s.add_dependency 'multi_string_replace', '~> 2.0'
   s.add_dependency 'nokogiri', '~> 1.15'
 
   s.metadata['rubygems_mfa_required'] = 'false'

data/spec/lib/pennmarc/heading_control_spec.rb

@@ -0,0 +1,45 @@
+# frozen_string_literal: true
+
+describe 'PennMARC::HeadingControl' do
+  let(:replace_term) { PennMARC::Mappers.heading_overrides.first[0] }
+  let(:replaced_term) { PennMARC::Mappers.heading_overrides.first[1] }
+  let(:remove_term) { PennMARC::Mappers.headings_to_remove.first }
+
+  describe '.process' do
+    context 'with a term for removal' do
+      it 'removes the term if found in isolation' do
+        values = [remove_term]
+        expect(PennMARC::HeadingControl.term_override(values)).to eq []
+      end
+
+      it 'removes the term regardless of case' do
+        values = [remove_term.downcase]
+        expect(PennMARC::HeadingControl.term_override(values)).to eq []
+      end
+
+      it 'removes the term if it is included as a substring' do
+        values = ["#{remove_term}--History"]
+        expect(PennMARC::HeadingControl.term_override(values)).to eq []
+      end
+    end
+
+    context 'with a term for replacement' do
+      it 'replaces the term in isolation' do
+        values = [replace_term]
+        expect(PennMARC::HeadingControl.term_override(values)).to eq [replaced_term]
+      end
+
+      it 'replaces the term when used with other headings' do
+        values = ["#{replace_term}--History"]
+        expect(PennMARC::HeadingControl.term_override(values)).to eq ["#{replaced_term}--History"]
+      end
+    end
+
+    context 'with a variety of terms' do
+      it 'removes and replaces terms as needed' do
+        values = [remove_term, replace_term, 'History']
+        expect(PennMARC::HeadingControl.term_override(values)).to contain_exactly 'History', replaced_term
+      end
+    end
+  end
+end

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

@@ -161,16 +161,36 @@ describe 'PennMARC::Identifier' do
   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'
+    context 'with a lower case tag' do
+      let(:record) do
+        marc_record fields: [
+          marc_field(tag: PennMARC::Enriched::Pub::RELATED_RECORD_TAGS.second, subfields: { w: '123456789',
+                                                                                            c: 'Contains',
+                                                                                            a: 'Title' }),
+          marc_field(tag: PennMARC::Enriched::Pub::RELATED_RECORD_TAGS.second, 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
+
+    context 'with an upper case tag' do
+      let(:record) do
+        marc_record fields: [
+          marc_field(tag: PennMARC::Enriched::Pub::RELATED_RECORD_TAGS.first, subfields: { w: '123456789',
+                                                                                           c: 'Contains',
+                                                                                           a: 'Title' }),
+          marc_field(tag: PennMARC::Enriched::Pub::RELATED_RECORD_TAGS.first, 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
 end

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

@@ -302,6 +302,25 @@ describe 'PennMARC::Subject' do
         expect(values).to contain_exactly 'Philosophy in motion pictures.'
       end
     end
+
+    context 'with headings that contain terms for removal and replacement' do
+      let(:fields) do
+        [marc_field(tag: '650', subfields: { a: 'History.' }),
+         marc_field(tag: '650', subfields: { a: PennMARC::Mappers.headings_to_remove.first }),
+         marc_field(tag: '650', subfields: { a: PennMARC::Mappers.heading_overrides.first[0] })]
+      end
+
+      it 'removes and replaces terms as expected' do
+        expect(values).to contain_exactly 'History.', "#{PennMARC::Mappers.heading_overrides.first[1]}."
+      end
+
+      it 'does not remove or replace terms if override param is false' do
+        expect(helper.show(record, override: false)).to contain_exactly(
+          'History.', "#{PennMARC::Mappers.headings_to_remove.first}.",
+          "#{PennMARC::Mappers.heading_overrides.first[0]}."
+        )
+      end
+    end
   end
 
   describe '.childrens_show' do

metadata

@@ -1,16 +1,18 @@
 --- !ruby/object:Gem::Specification
 name: pennmarc
 version: !ruby/object:Gem::Version
-  version: 1.0.26
+  version: 1.0.30
 platform: ruby
 authors:
 - Mike Kanning
 - Amrey Mathurin
 - Patrick Perkins
+- Katherine Schultz
+- Baowei Wei
 autorequire:
 bindir: bin
 cert_chain: []
-date: 2024-07-02 00:00:00.000000000 Z
+date: 2024-07-15 00:00:00.000000000 Z
 dependencies:
 - !ruby/object:Gem::Dependency
   name: activesupport
@@ -54,6 +56,20 @@ dependencies:
     - - "~>"
       - !ruby/object:Gem::Version
         version: '1.2'
+- !ruby/object:Gem::Dependency
+  name: multi_string_replace
+  requirement: !ruby/object:Gem::Requirement
+    requirements:
+    - - "~>"
+      - !ruby/object:Gem::Version
+        version: '2.0'
+  type: :runtime
+  prerelease: false
+  version_requirements: !ruby/object:Gem::Requirement
+    requirements:
+    - - "~>"
+      - !ruby/object:Gem::Version
+        version: '2.0'
 - !ruby/object:Gem::Dependency
   name: nokogiri
   requirement: !ruby/object:Gem::Requirement
@@ -116,6 +132,8 @@ files:
 - lib/pennmarc/helpers/title.rb
 - lib/pennmarc/mappers.rb
 - lib/pennmarc/mappings/dewey_classification.yml
+- lib/pennmarc/mappings/headings_override.yml
+- lib/pennmarc/mappings/headings_remove.yml
 - lib/pennmarc/mappings/iso639-2-languages.yml
 - lib/pennmarc/mappings/iso639-3-languages.yml
 - lib/pennmarc/mappings/loc_classification.yml
@@ -128,6 +146,7 @@ files:
 - lib/pennmarc/version.rb
 - pennmarc.gemspec
 - spec/fixtures/marcxml/test.xml
+- spec/lib/pennmarc/heading_control_spec.rb
 - spec/lib/pennmarc/helpers/access_spec.rb
 - spec/lib/pennmarc/helpers/citation_spec.rb
 - spec/lib/pennmarc/helpers/classification_spec.rb