pennmarc 1.3.5 → 1.4.0

checksums.yaml

@@ -1,7 +1,7 @@
 ---
 SHA256:
-  metadata.gz: 1d1e26fab095596cdef4b9e0cf3b8f0c571ce7526b5bf8a8b17b6b1240d1c2b9
-  data.tar.gz: 133829f994db6bc1ee7f341e6750c4202126dcc0dbe5bdaec3610170317164ab
+  metadata.gz: ea8b17f4fa038de9e5c44d384b82dfce5cc7a6cb909a5dc63fa1080351096e69
+  data.tar.gz: 58c48f68947d697689499018ca2abdc89d791db97e0ee076b74a12bf161d4470
 SHA512:
-  metadata.gz: 2616ca6b451b3dc99caef45e63d444b8e1d4b3d87e8cac920f68ec54bcddfe6193392e34d6efe23f9144173baa0fc271ff0b94b2f1f7eb7676d5ee0beb8d800a
-  data.tar.gz: f0a69e601d112df2b53b1107ff33e421d0d9da92845601bc24ed96980a6876bf56c79c25d3301f412d1dc20815ed591875aed0248957bff4c9179b14aa263cfd
+  metadata.gz: dad1e73ee44fa2635d81c6e0f1b7abc54d648b81679298121e6358c92f6872bf2f22d87f493005dad17b94402166cecb5c4751b3da33ae3a8ca35d90bc283e56
+  data.tar.gz: a23a12db492f9449632cfba5c7d1d247ab8302ede1094ac12ed0ff175f796787a2fa99dc9b17672d35ff67747523d9b4e40b31be7b602d910ae9a220d8f572a3

data/.pipeline.yml

@@ -1,3 +1,7 @@
+variables:
+  FF_SCRIPT_SECTIONS: "true"
+  GIT_CLONE_PATH: "${CI_BUILDS_DIR}/${CI_PROJECT_NAME}/${CI_JOB_ID}"
+
 include:
   - component: gitlab.library.upenn.edu/devops/gitlab/components/general/install_hashicorp_vault@~latest
   - component: gitlab.library.upenn.edu/devops/gitlab/components/general/vault_jwt_auth@~latest

data/.rubocop_todo.yml

@@ -1,12 +1,12 @@
 # This configuration was generated by
 # `rubocop --auto-gen-config --auto-gen-only-exclude --exclude-limit 10000`
-# on 2025-08-15 20:13:24 UTC using RuboCop version 1.79.2.
+# on 2025-10-22 21:06:47 UTC using RuboCop version 1.79.2.
 # The point is for the user to remove these configuration records
 # one by one as the offenses are removed from the code base.
 # Note that changes in the inspected code, or installation of new
 # versions of RuboCop, may require this file to be generated again.
 
-# Offense count: 27
+# Offense count: 28
 # Configuration parameters: AllowedMethods, AllowedPatterns, CountRepeatedAttributes, Max.
 Metrics/AbcSize:
   Exclude:
@@ -33,7 +33,7 @@ Metrics/ClassLength:
     - 'lib/pennmarc/helpers/subject.rb'
     - 'lib/pennmarc/helpers/title.rb'
 
-# Offense count: 21
+# Offense count: 22
 # Configuration parameters: AllowedMethods, AllowedPatterns, Max.
 Metrics/CyclomaticComplexity:
   Exclude:
@@ -50,7 +50,7 @@ Metrics/CyclomaticComplexity:
     - 'lib/pennmarc/helpers/title.rb'
     - 'lib/pennmarc/util.rb'
 
-# Offense count: 17
+# Offense count: 18
 # Configuration parameters: CountComments, Max, CountAsOne, AllowedMethods, AllowedPatterns.
 Metrics/MethodLength:
   Exclude:
@@ -70,7 +70,7 @@ Metrics/ModuleLength:
   Exclude:
     - 'lib/pennmarc/util.rb'
 
-# Offense count: 16
+# Offense count: 17
 # Configuration parameters: AllowedMethods, AllowedPatterns, Max.
 Metrics/PerceivedComplexity:
   Exclude:
@@ -109,9 +109,18 @@ RSpec/ExampleLength:
   Exclude:
     - 'spec/lib/pennmarc/marc_util_spec.rb'
 
-# Offense count: 1
+# Offense count: 2
 # Configuration parameters: Include, CustomTransform, IgnoreMethods, IgnoreMetadata.
 # Include: **/*_spec.rb
 RSpec/SpecFilePathFormat:
   Exclude:
     - 'spec/lib/pennmarc/parser_spec.rb'
+    - 'spec/lib/pennmarc/services/title_suggestion_weight_service_spec.rb'
+
+# Offense count: 1
+# This cop supports unsafe autocorrection (--autocorrect-all).
+# Configuration parameters: EnforcedStyle, AllowedMethods, AllowedPatterns.
+# SupportedStyles: predicate, comparison
+Style/NumericPredicate:
+  Exclude:
+    - 'lib/pennmarc/services/title_suggestion_weight_service.rb'

data/Gemfile

@@ -19,6 +19,7 @@ group :test do
 end
 
 group :development do
-  gem 'webrick', '~> 1.8'
+  gem 'puma'
+  gem 'rackup'
   gem 'yard', '~> 0.9'
 end

data/Gemfile.lock

@@ -35,6 +35,7 @@ GEM
       nokogiri (~> 1.0)
       rexml
     minitest (5.26.0)
+    nio4r (2.7.4)
     nokogiri (1.18.9-arm64-darwin)
       racc (~> 1.4)
     nokogiri (1.18.9-x64-mingw-ucrt)
@@ -48,8 +49,12 @@ GEM
       ast (~> 2.4.1)
       racc
     prism (1.4.0)
+    puma (7.1.0)
+      nio4r (~> 2.0)
     racc (1.8.1)
-    rack (3.2.0)
+    rack (3.2.3)
+    rackup (2.2.1)
+      rack (>= 3)
     rainbow (3.1.1)
     rake (13.3.0)
     regexp_parser (2.11.2)
@@ -121,7 +126,6 @@ GEM
       rubocop-rake
       rubocop-rspec
     uri (1.0.4)
-    webrick (1.8.1)
     yard (0.9.37)
 
 PLATFORMS
@@ -137,11 +141,12 @@ DEPENDENCIES
   library_stdnums (~> 1.6)
   marc (~> 1.2)
   nokogiri (~> 1.15)
+  puma
+  rackup
   rake (~> 13.0)
   rspec (~> 3.12)
   simplecov (~> 0.22)
   upennlib-rubocop
-  webrick (~> 1.8)
   yard (~> 0.9)
 
 BUNDLED WITH

data/lib/pennmarc/helpers/date.rb

@@ -36,7 +36,7 @@ module PennMARC
 
             Time.strptime(date_added, format)
           rescue StandardError => e
-            puts 'Error parsing date in date added subfield. ' \
+            warn 'Error parsing date in date added subfield. ' \
                  "mmsid: #{Identifier.mmsid(record)}, value: #{date_added}, error: #{e}"
             nil
           end
@@ -59,7 +59,7 @@ module PennMARC
 
             Time.strptime(date_time_string, '%Y%m%d%H%M%S.%N')
           rescue StandardError => e
-            puts 'Error parsing last updated date. ' \
+            warn 'Error parsing last updated date. ' \
                  "mmsid: #{Identifier.mmsid(record)}, value: #{date_time_string}, error: #{e}"
             nil
           end

data/lib/pennmarc/helpers/title.rb

@@ -1,5 +1,7 @@
 # frozen_string_literal: true
 
+require_relative '../services/title_suggestion_weight_service'
+
 module PennMARC
   # This helper contains logic for parsing out Title and Title-related fields.
   class Title < Helper
@@ -36,6 +38,27 @@ module PennMARC
     NO_TITLE_PROVIDED = '[No title provided]'
 
     class << self
+      # Values for title suggester, including only ǂa and ǂb from
+      # {https://www.loc.gov/marc/bibliographic/bd245.html 245} field. Limits the output to 20 words and strips any
+      # trailing slashes.
+      # @param record [MARC::Record]
+      # @return [Array<String>] array of all title values for suggestion
+      def suggest(record)
+        record.fields(%w[245]).filter_map do |field|
+          join_subfields(field, &subfield_in?(%w[a b]))
+            .squish
+            .truncate_words(20)
+            .sub(%r{ /$}, '')
+        end
+      end
+
+      # An integer value used for weighing title suggest values. See {PennMARC::TitleSuggestionWeightService} for logic.
+      # @param record [MARC::Record]
+      # @return [Integer]
+      def suggest_weight(record)
+        PennMARC::TitleSuggestionWeightService.weight record
+      end
+
       # 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 record [MARC::Record]

data/lib/pennmarc/mappers.rb

@@ -50,10 +50,9 @@ module PennMARC
       end
 
       # @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
+      # @param symbolize_names [Boolean] whether 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: symbolize_names)
       end

data/lib/pennmarc/services/title_suggestion_weight_service.rb

@@ -0,0 +1,102 @@
+# frozen_string_literal: true
+
+module PennMARC
+  # A service to calculate suggestion weights based on a variety of criteria
+  class TitleSuggestionWeightService
+    # Starting score
+    BASE_WEIGHT = 10
+
+    # Array of symbols referring to methods on this object that return a boolean and the scoring factor if the
+    # method returns true.
+    FACTORS = [
+      [:targeted_format?,             8],
+      [:published_in_last_ten_years?, 5],
+      [:electronic_holdings?,         3],
+      [:high_encoding_level?,         2],
+      [:physical_holdings?,           1],
+      [:low_encoding_level?,         -2],
+      [:weird_format?,               -5],
+      [:no_holdings?,                -10]
+    ].freeze
+
+    # Score higher records with these formats
+    TARGETED_FORMATS = [Format::BOOK, Format::WEBSITE_DATABASE, Format::JOURNAL_PERIODICAL, Format::NEWSPAPER,
+                        Format::SOUND_RECORDING, Format::MUSICAL_SCORE].freeze
+    # Score lower these formats
+    WEIRD_FORMATS = [Format::OTHER, Format::THREE_D_OBJECT].freeze
+
+    # See #{PennMARC::EncodingLevel} for more of the logic that determines sort values
+    # An encoding sort level of this value is considered good
+    HIGH_ENCODING_SORT_LEVEL = 0
+    # An encoding sort level higher than this is considered poor
+    LOW_ENCODING_SORT_LEVEL = 4
+
+    class << self
+      # Calculate a weight for use in sorting good title suggestions from bad
+      # @param record [MARC::Record]
+      # @return [Integer]
+      def weight(record)
+        factors.reduce(BASE_WEIGHT) do |weight, (call, score)|
+          weight + (public_send(call, record) ? score : 0)
+        end
+      end
+
+      # @return [Array[Array]]
+      def factors
+        FACTORS
+      end
+
+      # @param record [MARC::Record]
+      # @return [Boolean]
+      def published_in_last_ten_years?(record)
+        return false unless Date.publication(record).present?
+
+        Date.publication(record) > 10.years.ago
+      end
+
+      # @param record [MARC::Record]
+      # @return [Boolean]
+      def electronic_holdings?(record)
+        Inventory.electronic(record)&.any? || false
+      end
+
+      # @param record [MARC::Record]
+      # @return [Boolean]
+      def physical_holdings?(record)
+        Inventory.physical(record)&.any? || false
+      end
+
+      # @param record [MARC::Record]
+      # @return [Boolean]
+      def targeted_format?(record)
+        (Format.facet(record) & TARGETED_FORMATS).any?
+      end
+
+      # @param record [MARC::Record]
+      # @return [Boolean]
+      def high_encoding_level?(record)
+        Encoding.level_sort(record) == HIGH_ENCODING_SORT_LEVEL
+      end
+
+      # @param record [MARC::Record]
+      # @return [Boolean]
+      def weird_format?(record)
+        (Format.facet(record) & WEIRD_FORMATS).any?
+      end
+
+      # @param record [MARC::Record]
+      # @return [Boolean]
+      def no_holdings?(record)
+        !electronic_holdings?(record) && !physical_holdings?(record)
+      end
+
+      # @param record [MARC::Record]
+      # @return [Boolean]
+      def low_encoding_level?(record)
+        return false unless Encoding.level_sort(record).present?
+
+        Encoding.level_sort(record) > LOW_ENCODING_SORT_LEVEL
+      end
+    end
+  end
+end

data/lib/pennmarc/version.rb

@@ -1,5 +1,5 @@
 # frozen_string_literal: true
 
 module PennMARC
-  VERSION = '1.3.5'
+  VERSION = '1.4.0'
 end

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

@@ -90,7 +90,7 @@ describe 'PennMARC::Date' do
         expect {
           helper.added(record)
         }.to output('Error parsing date in date added subfield. mmsid: mmsid, value: invalid date, ' \
-                    "error: invalid date or strptime format - `invalid date' `%Y-%m-%d %H:%M:%S'\n").to_stdout
+                    "error: invalid date or strptime format - `invalid date' `%Y-%m-%d %H:%M:%S'\n").to_stderr
       end
     end
   end
@@ -120,7 +120,7 @@ describe 'PennMARC::Date' do
         expect {
           helper.last_updated(record)
         }.to output('Error parsing last updated date. mmsid: mmsid, value: invalid date, ' \
-                    "error: invalid date or strptime format - `invalid date' `%Y%m%d%H%M%S.%N'\n").to_stdout
+                    "error: invalid date or strptime format - `invalid date' `%Y%m%d%H%M%S.%N'\n").to_stderr
       end
     end
   end

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

@@ -6,6 +6,39 @@ describe 'PennMARC::Title' do
   let(:fields) { [marc_field(tag: '245', subfields: subfields)] }
   let(:record) { marc_record fields: fields, leader: leader }
 
+  describe '.suggest' do
+    context 'with slashes in the title as well as at the end of ǂb' do
+      let(:subfields) { { a: 'Title /', b: 'Subtitle /' } }
+
+      it 'removes only the trailing slash' do
+        expect(helper.suggest(record).first).to eq 'Title / Subtitle'
+      end
+    end
+
+    context 'with a long title more than twenty words' do
+      let(:subfields) do
+        { a: 'The Book of Very Short Words With a Lot of Words',
+          b: 'Containing many words with that may or may not have many characters /' }
+      end
+
+      it 'truncates the title to twenty words and adds a trailing ellipsis' do
+        truncated_suggestion = helper.suggest(record).first
+        expect(truncated_suggestion).to eq(
+          'The Book of Very Short Words With a Lot of Words Containing many words with that may or may not...'
+        )
+        expect(truncated_suggestion.split(' ').count).to eq 20
+      end
+    end
+
+    context 'with other subfields present' do
+      let(:subfields) { { a: 'Title', b: 'Subtitle', c: 'Author' } }
+
+      it 'returns only title fields' do
+        expect(helper.suggest(record).first).not_to include 'Author'
+      end
+    end
+  end
+
   describe '.search' do
     let(:fields) do
       [marc_field(tag: '245', subfields: { a: 'Title', b: 'Subtitle', c: 'Responsibility', h: 'Medium' }),

data/spec/lib/pennmarc/services/title_suggestion_weight_service_spec.rb

@@ -0,0 +1,231 @@
+# frozen_string_literal: true
+
+describe PennMARC::TitleSuggestionWeightService do
+  let(:record) { instance_double MARC::Record }
+
+  describe '.weight' do
+    context 'with defined factors' do
+      before do
+        allow(described_class).to receive_messages(
+          targeted_format?: true,
+          published_in_last_ten_years?: false,
+          electronic_holdings?: false,
+          high_encoding_level?: false,
+          physical_holdings?: false,
+          low_encoding_level?: false,
+          weird_format?: false,
+          no_holdings?: false
+        )
+      end
+
+      it 'scores properly based on factor responce valence' do
+        expected_score = described_class::BASE_WEIGHT + described_class::FACTORS[0].second
+        expect(described_class.weight(record)).to eq expected_score
+      end
+    end
+  end
+
+  describe '.published_in_the_last_ten_years' do
+    before { allow(PennMARC::Date).to receive(:publication).with(record).and_return(record_date) }
+
+    let(:value) { described_class.published_in_last_ten_years?(record) }
+
+    context 'with no date' do
+      let(:record_date) { nil }
+
+      it 'returns true' do
+        expect(value).to be false
+      end
+    end
+
+    context 'with a recent date' do
+      let(:record_date) { Time.now }
+
+      it 'returns true' do
+        expect(value).to be true
+      end
+    end
+
+    context 'with an ancient date' do
+      let(:record_date) { Time.now - 400.years }
+
+      it 'returns false' do
+        expect(value).to be false
+      end
+    end
+  end
+
+  describe '.targeted_format?' do
+    before { allow(PennMARC::Format).to receive(:facet).with(record).and_return([record_format]) }
+
+    let(:value) { described_class.targeted_format?(record) }
+
+    context 'with no format' do
+      let(:record_format) { nil }
+
+      it 'returns false' do
+        expect(value).to be false
+      end
+    end
+
+    context 'with a targeted format' do
+      let(:record_format) { PennMARC::TitleSuggestionWeightService::TARGETED_FORMATS.sample }
+
+      it 'returns true' do
+        expect(value).to be true
+      end
+    end
+
+    context 'with a non-targeted format' do
+      let(:record_format) { PennMARC::TitleSuggestionWeightService::WEIRD_FORMATS.sample }
+
+      it 'returns false' do
+        expect(value).to be false
+      end
+    end
+  end
+
+  describe '.weird_format?' do
+    before { allow(PennMARC::Format).to receive(:facet).with(record).and_return([record_format]) }
+
+    let(:value) { described_class.weird_format?(record) }
+
+    context 'with no format' do
+      let(:record_format) { nil }
+
+      it 'returns false' do
+        expect(value).to be false
+      end
+    end
+
+    context 'with a weird format' do
+      let(:record_format) { PennMARC::TitleSuggestionWeightService::WEIRD_FORMATS.sample }
+
+      it 'returns true' do
+        expect(value).to be true
+      end
+    end
+
+    context 'with a non-weird format' do
+      let(:record_format) { PennMARC::TitleSuggestionWeightService::TARGETED_FORMATS.sample }
+
+      it 'returns false' do
+        expect(value).to be false
+      end
+    end
+  end
+
+  describe '.low_encoding_level?' do
+    before { allow(PennMARC::Encoding).to receive(:level_sort).with(record).and_return(encoding_sort_score) }
+
+    let(:value) { described_class.low_encoding_level?(record) }
+
+    context 'with no encoding level' do
+      let(:encoding_sort_score) { nil }
+
+      it 'returns false' do
+        expect(value).to be false
+      end
+    end
+
+    context 'with a low encoding level' do
+      let(:encoding_sort_score) { 11 }
+
+      it 'returns true' do
+        expect(value).to be true
+      end
+    end
+
+    context 'with a high encoding level' do
+      let(:encoding_sort_score) { PennMARC::TitleSuggestionWeightService::HIGH_ENCODING_SORT_LEVEL }
+
+      it 'returns false' do
+        expect(value).to be false
+      end
+    end
+  end
+
+  describe '.electronic_holdings?' do
+    before do
+      allow(PennMARC::Inventory).to receive(:electronic).with(record).and_return(holdings)
+    end
+
+    let(:value) { described_class.electronic_holdings?(record) }
+
+    context 'with electronic holdings' do
+      let(:holdings) { [PennMARC::InventoryEntry::Electronic] }
+
+      it 'returns true' do
+        expect(value).to be true
+      end
+    end
+
+    context 'without any holdings' do
+      let(:holdings) { [] }
+
+      it 'returns false' do
+        expect(value).to be false
+      end
+    end
+  end
+
+  describe '.physical_holdings?' do
+    before do
+      allow(PennMARC::Inventory).to receive(:physical).with(record).and_return(holdings)
+    end
+
+    let(:value) { described_class.physical_holdings?(record) }
+
+    context 'with physical holdings' do
+      let(:holdings) { [PennMARC::InventoryEntry::Physical] }
+
+      it 'returns true' do
+        expect(value).to be true
+      end
+    end
+
+    context 'without any holdings' do
+      let(:holdings) { [] }
+
+      it 'returns false' do
+        expect(value).to be false
+      end
+    end
+  end
+
+  describe '.no_holdings?' do
+    before do
+      allow(PennMARC::Inventory).to receive(:physical).with(record).and_return(physical_holdings)
+      allow(PennMARC::Inventory).to receive(:electronic).with(record).and_return(electronic_holdings)
+    end
+
+    let(:value) { described_class.no_holdings?(record) }
+
+    context 'with neither physical nor electronic holdings' do
+      let(:physical_holdings) { [] }
+      let(:electronic_holdings) { [] }
+
+      it 'returns true' do
+        expect(value).to be true
+      end
+    end
+
+    context 'with only electronic holdings' do
+      let(:physical_holdings) { [] }
+      let(:electronic_holdings) { [instance_double(PennMARC::InventoryEntry::Electronic)] }
+
+      it 'returns false' do
+        expect(value).to be false
+      end
+    end
+
+    context 'with only physical holdings' do
+      let(:physical_holdings) { [instance_double(PennMARC::InventoryEntry::Physical)] }
+      let(:electronic_holdings) { [] }
+
+      it 'returns false' do
+        expect(value).to be false
+      end
+    end
+  end
+end

metadata

@@ -1,7 +1,7 @@
 --- !ruby/object:Gem::Specification
 name: pennmarc
 version: !ruby/object:Gem::Version
-  version: 1.3.5
+  version: 1.4.0
 platform: ruby
 authors:
 - Mike Kanning
@@ -12,7 +12,7 @@ authors:
 autorequire:
 bindir: bin
 cert_chain: []
-date: 2025-10-16 00:00:00.000000000 Z
+date: 2025-10-31 00:00:00.000000000 Z
 dependencies:
 - !ruby/object:Gem::Dependency
   name: activesupport
@@ -143,6 +143,7 @@ files:
 - lib/pennmarc/mappings/locations.yml
 - lib/pennmarc/mappings/relator.yml
 - lib/pennmarc/parser.rb
+- lib/pennmarc/services/title_suggestion_weight_service.rb
 - lib/pennmarc/test/marc_helpers.rb
 - lib/pennmarc/util.rb
 - lib/pennmarc/version.rb
@@ -172,6 +173,7 @@ files:
 - spec/lib/pennmarc/helpers/title_spec.rb
 - spec/lib/pennmarc/marc_util_spec.rb
 - spec/lib/pennmarc/parser_spec.rb
+- spec/lib/pennmarc/services/title_suggestion_weight_service_spec.rb
 - spec/spec_helper.rb
 - spec/support/fixture_helpers.rb
 homepage: https://gitlab.library.upenn.edu/dld/catalog/pennmarc