pennmarc 0.0.2 → 1.0.1

data/lib/pennmarc/helpers/subject.rb

@@ -26,14 +26,14 @@ module PennMARC
       # Local subject heading tags
       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}.
+      # 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}.
       # @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
       # @param [MARC::Record] record
       # @return [Array] array of all subject values for search
-      def search(record, relator_map)
+      def search(record, relator_map: Mappers.relator)
         subject_fields(record, type: :search).filter_map do |field|
           subj_parts = field.filter_map do |subfield|
             # TODO: use term hash here? pro/chr would be rejected...
@@ -45,10 +45,9 @@ module PennMARC
               # remove any ? at the end
               subfield.value.gsub(/^%?(PRO|CHR)/, '').gsub(/\?$/, '').strip
             when '4'
-              # TODO: use relation mapping method from Title helper? for potential URI support?
               # sf 4 should contain a 3-letter code or URI "that specifies the relationship from the entity described
               # in the record to the entity referenced in the field"
-              "#{subfield.value} #{relator_map[subfield.value.to_sym]}".strip
+              "#{subfield.value} #{translate_relator(subfield.value.to_sym, relator_map)}".strip
             else
               subfield.value
             end
@@ -79,12 +78,12 @@ module PennMARC
       # @param [MARC::Record] record
       # @return [Array] array of all subject values for display
       def show(record)
-        subject_fields(record, type: :all).filter_map do |field|
+        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
-        end.uniq
+        }.uniq
       end
 
       # Get Subjects from "Children" ontology
@@ -93,12 +92,12 @@ module PennMARC
       # @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 do |field|
+          .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
-          end.uniq
+          }.uniq
       end
 
       # Get Subjects from "MeSH" ontology
@@ -107,12 +106,12 @@ module PennMARC
       # @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 do |field|
+          .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
-          end.uniq
+          }.uniq
       end
 
       # Get Subject values from {DISPLAY_TAGS} where indicator2 is 4 and {LOCAL_TAGS}. Do not include any values where
@@ -123,14 +122,14 @@ module PennMARC
       def local_show(record)
         local_fields = subject_fields(record, type: :display, options: { tags: DISPLAY_TAGS, indicator2: '4' }) +
                        subject_fields(record, type: :local)
-        local_fields.filter_map do |field|
+        local_fields.filter_map { |field|
           next if subfield_value?(field, '2', /penncoi/)
 
           term_hash = build_subject_hash(field)
           next if term_hash.blank? || term_hash[:count]&.zero?
 
           format_term type: :display, term: term_hash
-        end.uniq
+        }.uniq
       end
 
       private
@@ -183,7 +182,7 @@ module PennMARC
       # @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
+        return false unless field.tag.in?(DISPLAY_TAGS + LOCAL_TAGS) && field.respond_to?(:indicator2)
 
         return false if field.indicator2 == '7' && !valid_subject_genre_source_code?(field)
 
@@ -200,7 +199,7 @@ module PennMARC
       # @param [Hash] options include :tags and :indicator2 values
       # @return [Boolean] whether a MARC field should be considered for display
       def subject_display_field?(field, options)
-        return false if field.blank?
+        return false unless field.respond_to?(:indicator2)
 
         return true if field.tag.in?(options[:tags]) && field.indicator2.in?(options[:indicator2])
 
@@ -210,7 +209,7 @@ module PennMARC
       # @param [MARC::DataField] field
       # @return [Boolean]
       def subject_facet_field?(field)
-        return false if field.blank?
+        return false unless field.respond_to?(:indicator2)
 
         return true if field.tag.in?(DISPLAY_TAGS) && field.indicator2.in?(%w[0 2 4])
 
@@ -269,7 +268,7 @@ module PennMARC
       # @param [MARC::DataField] field
       # @return [Boolean]
       def subject_search_field?(field)
-        return false if field.blank? || SEARCH_SOURCE_INDICATORS.exclude?(field.indicator2)
+        return false unless field.respond_to?(:indicator2) && SEARCH_SOURCE_INDICATORS.include?(field.indicator2)
 
         tag = if field.tag == '880'
                 subfield_values(field, '6').first

data/lib/pennmarc/helpers/title.rb

@@ -82,7 +82,7 @@ module PennMARC
       def sort(record)
         title_field = record.fields('245').first
         # attempt to get number of non-filing characters present, default to 0
-        offset = if title_field.indicator2 =~ /^[0-9]$/
+        offset = if /^[0-9]$/.match?(title_field.indicator2)
                    title_field.indicator2.to_i
                  else
                    0

data/lib/pennmarc/mappers.rb

@@ -0,0 +1,31 @@
+# frozen_string_literal: true
+
+module PennMARC
+  # reusable static mappers
+  class Mappers
+    class << self
+      # @return [Hash]
+      def language
+        @language ||= load_map('language.yml')
+      end
+
+      # @return [Hash]
+      def location
+        @location ||= load_map('locations.yml')
+      end
+
+      # @return [Hash]
+      def relator
+        @relator ||= load_map('relator.yml')
+      end
+
+      # @param [String] filename of mapping file in config directory, with file extension
+      # @return [Hash] mapping as hash
+      def load_map(filename)
+        puts { "Loading #{filename}" }
+        YAML.safe_load(File.read(File.join(File.expand_path(__dir__), 'mappings', filename)),
+                       symbolize_names: true)
+      end
+    end
+  end
+end

data/lib/pennmarc/parser.rb

@@ -2,176 +2,53 @@
 
 require 'active_support/all'
 require_relative 'helpers/helper'
-require_relative 'helpers/creator'
-require_relative 'helpers/database'
-require_relative 'helpers/date'
-require_relative 'helpers/format'
-require_relative 'helpers/genre'
-require_relative 'helpers/identifier'
-require_relative 'helpers/language'
-require_relative 'helpers/link'
-require_relative 'helpers/location'
-require_relative 'helpers/subject'
-require_relative 'helpers/title'
-require_relative 'helpers/citation'
-require_relative 'helpers/relation'
-require_relative 'helpers/production'
-require_relative 'helpers/edition'
-require_relative 'helpers/note'
 
-module PennMARC
-  attr_accessor :mappings
-
-  DEFINED_HELPERS = %w[Creator Database Date Format Genre Language Link Location Subject Title Relation].freeze
+# Require all files in helpers directory
+# TODO: this double-requires Helper, but that needs to be required before other helpers...
+Dir[File.join(__dir__, 'helpers', '*.rb')].each { |file| require file }
 
-  # Methods here should return values used in the indexer. The parsing logic should
-  # NOT return values specific to any particular site/interface, but just general
-  # MARC parsing logic for "title", "subject", "author", etc., as much as reasonably
-  # possible. We'll see how it goes.
-  #
-  # Methods should, by default, take in a MARC::Record
+# Top level gem namespace
+module PennMARC
+  # Access point for magic calls to helper methods
   class Parser
-    def initialize(helpers: DEFINED_HELPERS)
-      @mappings = {}
-      @helpers = Array.wrap(helpers) # TODO: load helpers dynamically?
-    end
-
-    def respond_to_missing?(name)
-      name.split('_').first.in? @helpers
+    # Allow calls to `respond_to?` on parser instances to respond accurately by checking helper classes
+    # @param [String, Symbol] name
+    # @return [Boolean]
+    def respond_to_missing?(name, include_private = false)
+      helper, method_name = parse_call(name)
+      begin
+        "PennMARC::#{helper}".constantize.respond_to?(method_name)
+      rescue NameError
+        super # Helper is not defined, so check self
+      end
     end
 
-    # Call helper class methods, e.g.,
+    # Call helper class methods, passing along additional arguments as needed, e.g.:
     # #title_show -> PennMARC::Title.show
     # #subject_facet -> PennMARC::Subject.facet
-    def method_missing(name, opts)
-      call = name.to_s.split('_')
-      helper = call.shift
-      meth = call.join('_')
-      "PennMARC::#{helper.titleize}".constantize.public_send(meth, opts)
-    end
-
-    # @todo does this fit in an existing helper?
+    # @param [Symbol] name
     # @param [MARC::Record] record
-    # @return [Object]
-    def cartographic_show(record)
-      record.fields(%w{255 342}).map do |field|
-        join_subfields(field, &subfield_not_6_or_8)
-      end
-    end
+    # @param [Array] opts
+    def method_missing(name, record, *opts)
+      helper, method_name = parse_call(name)
+      raise NoMethodError unless helper && method_name
 
-    # @todo move to Identifier helper
-    # @param [MARC::Record] record
-    # @return [Object]
-    def fingerprint_show(record)
-      record.fields('026').map do |field|
-        join_subfields(field, &subfield_not_in(%w{2 5 6 8}))
+      helper_klass = "PennMARC::#{helper.titleize}".constantize
+      if opts.any?
+        helper_klass.public_send(method_name, record, **opts.first)
+      else
+        helper_klass.public_send(method_name, record)
       end
     end
 
-    # @todo does this fit in an existing helper?
-    # @param [MARC::Record] record
-    # @return [Object]
-    def arrangement_show(record)
-      get_datafield_and_880(record, '351')
-    end
+    private
 
-    # @param [MARC::Record] record
-    # @return [Object]
-    def system_details_show(record)
-      acc = []
-      acc += record.fields('538').map do |field|
-        get_sub3_and_other_subs(field, &subfield_in(%w{a i u}))
-      end
-      acc += record.fields('344').map do |field|
-        get_sub3_and_other_subs(field, &subfield_in(%w{a b c d e f g h}))
-      end
-      acc += record.fields(%w{345 346}).map do |field|
-        get_sub3_and_other_subs(field, &subfield_in(%w{a b}))
-      end
-      acc += record.fields('347').map do |field|
-        get_sub3_and_other_subs(field, &subfield_in(%w{a b c d e f}))
-      end
-      acc += record.fields('880')
-                .select { |f| has_subfield6_value(f, /^538/) }
-                .map do |field|
-        get_sub3_and_other_subs(field, &subfield_in(%w{a i u}))
-      end
-      acc += record.fields('880')
-                .select { |f| has_subfield6_value(f, /^344/) }
-                .map do |field|
-        get_sub3_and_other_subs(field, &subfield_in(%w{a b c d e f g h}))
-      end
-      acc += record.fields('880')
-                .select { |f| has_subfield6_value(f, /^(345|346)/) }
-                .map do |field|
-        get_sub3_and_other_subs(field, &subfield_in(%w{a b}))
-      end
-      acc += record.fields('880')
-                .select { |f| has_subfield6_value(f, /^347/) }
-                .map do |field|
-        get_sub3_and_other_subs(field, &subfield_in(%w{a b c d e f}))
-      end
-      acc
-    end
-
-    # @todo the legacy code here is a hot mess for a number of reasons, what do we need this field to do?
-    # @note port the needed parts from get_offsite_display, don't return HTML
-    # @param [MARC::Record] record
-    # @return [Object]
-    def offsite_show(record); end
-
-    # @todo move this to Creator helper
-    # @param [MARC::Record] record
-    # @return [Object]
-    def contributor_show(record)
-      acc = []
-      acc += record.fields(%w{700 710})
-                   .select { |f| ['', ' ', '0'].member?(f.indicator2) }
-                   .select { |f| f.none? { |sf| sf.code == 'i' } }
-                   .map do |field|
-        contributor = join_subfields(field, &subfield_in(%w{a b c d j q}))
-        contributor_append = field.select(&subfield_in(%w{e u 3 4})).map do |sf|
-          if sf.code == '4'
-            ", #{relator_codes[sf.value]}"
-          else
-            " #{sf.value}"
-          end
-        end.join
-        { value: contributor, value_append: contributor_append, link_type: 'author_creator_xfacet2' }
-      end
-      acc += record.fields('880')
-                   .select { |f| has_subfield6_value(f, /^(700|710)/) && (f.none? { |sf| sf.code == 'i' }) }
-                   .map do |field|
-        contributor = join_subfields(field, &subfield_in(%w{a b c d j q}))
-        contributor_append = join_subfields(field, &subfield_in(%w{e u 3}))
-        { value: contributor, value_append: contributor_append, link_type: 'author_creator_xfacet2' }
-      end
-      acc
-    end
-
-    # Load language map from YAML and memoize in @mappings hash
-    # @return [Hash]
-    def language_map
-      @mappings[:language] ||= load_map('language.yml')
-    end
-
-    # Load location map from YAML and memoize in @mappings hash
-    # @return [Hash]
-    def location_map
-      @mappings[:location] ||= load_map('locations.yml')
-    end
-
-    # Load relator map from YAML and memoize in @mappings hash
-    # @return [Hash]
-    def relator_map
-      @mappings[:relator] ||= load_map('relator.yml')
-    end
-
-    # @param [String] filename of mapping file in config directory, with file extension
-    # @return [Hash] mapping as hash
-    def load_map(filename)
-      YAML.safe_load(File.read(File.join(File.expand_path(__dir__), 'mappings', filename)),
-                     symbolize_names: true)
+    # Parse out a method call name in the way method_missing is configured to handle
+    # @param [String, Symbol] name
+    # @return [Array]
+    def parse_call(name)
+      call = name.to_s.split('_')
+      [call.shift&.titleize, call.join('_').to_sym]
     end
   end
 end

data/lib/pennmarc/util.rb

@@ -10,9 +10,11 @@ module PennMARC
     # @param [Proc] selector
     # @return [String]
     def join_subfields(field, &selector)
-      field.select { |v| selector.call(v) }.filter_map { |sf|
+      return '' unless field
+
+      field.select(&selector).filter_map { |sf|
         value = sf.value&.strip
-        next unless value.present?
+        next if value.blank?
 
         value
       }.join(' ').squish
@@ -26,7 +28,7 @@ module PennMARC
     # @param [Regexp] regex
     # @return [TrueClass, FalseClass]
     def subfield_value?(field, subfield, regex)
-      field.any? { |sf| sf.code == subfield.to_s && sf.value =~ regex }
+      field&.any? { |sf| sf.code == subfield.to_s && sf.value =~ regex }
     end
 
     # returns true if a given field has a given subfield value in a given array
@@ -49,7 +51,6 @@ module PennMARC
     end
 
     # returns a lambda checking if passed-in subfield's code is a member of array
-    # TODO: include lambda returning methods in their own module?
     # @param [Array] array
     # @return [Proc]
     def subfield_in?(array)
@@ -57,7 +58,6 @@ module PennMARC
     end
 
     # returns a lambda checking if passed-in subfield's code is NOT a member of array
-    # TODO: include lambda returning methods in their own module?
     # @param [Array] array
     # @return [Proc]
     def subfield_not_in?(array)
@@ -88,7 +88,7 @@ module PennMARC
       field.filter_map do |sf|
         next unless sf.code == subfield.to_s
 
-        next unless sf.value.present?
+        next if sf.value.blank?
 
         sf.value
       end
@@ -124,16 +124,15 @@ module PennMARC
     # See: https://www.loc.gov/marc/bibliographic/bd880.html
     # @param [MARC::Record] record
     # @param [String|Array] subfield6_value either a string to look for in sub6 or an array of them
-    # @param selector [Proc] takes a subfield as argument, returns a boolean
+    # @param [Proc] selector takes a subfield as argument, returns a boolean
     # @return [Array] array of linked alternates
     def linked_alternate(record, subfield6_value, &selector)
       record.fields('880').filter_map do |field|
         next unless subfield_value?(field, '6', /^#{Array.wrap(subfield6_value).join('|')}/)
 
-        field.select { |sf| selector.call(sf) }.map(&:value).join(' ')
+        field.select(&selector).map(&:value).join(' ')
       end
     end
-    alias get_880 linked_alternate
 
     # Common case of wanting to extract all the subfields besides 6 or 8,
     # from 880 datafield that has a particular subfield 6 value. We exclude 6 because
@@ -142,8 +141,9 @@ module PennMARC
     # @param [String|Array] subfield6_value either a string to look for in sub6 or an array of them
     # @return [Array] array of linked alternates without 8 or 6 values
     def linked_alternate_not_6_or_8(record, subfield6_value)
+      excluded_subfields = %w[6 8]
       linked_alternate(record, subfield6_value) do |sf|
-        %w[6 8].exclude?(sf.code)
+        excluded_subfields.exclude?(sf.code)
       end
     end
 
@@ -152,9 +152,9 @@ module PennMARC
     # @param [String] tag
     # @return [Array] acc
     def datafield_and_linked_alternate(record, tag)
-      record.fields(tag).filter_map do |field|
+      record.fields(tag).filter_map { |field|
         join_subfields(field, &subfield_not_in?(%w[6 8]))
-      end + linked_alternate_not_6_or_8(record, tag)
+      } + linked_alternate_not_6_or_8(record, tag)
     end
 
     # Get the substring of a string up to a given target character
@@ -179,14 +179,13 @@ module PennMARC
     def join_and_squish(array)
       array.join(' ').squish
     end
-    alias join_and_trim_whitespace join_and_squish
 
     # If there's a subfield i, extract its value, and if there's something
     # in parentheses in that value, extract that.
     # @param [MARC::Field] field
     # @return [String] subfield i without parentheses value
     def remove_paren_value_from_subfield_i(field)
-      val = field.filter_map do |sf|
+      val = field.filter_map { |sf|
         next unless sf.code == 'i'
 
         match = /\((.+?)\)/.match(sf.value)
@@ -195,7 +194,7 @@ module PennMARC
         else
           sf.value
         end
-      end.first || ''
+      }.first || ''
       trim_trailing(:colon, trim_trailing(:period, val))
     end
 
@@ -205,7 +204,7 @@ module PennMARC
     # @param [Hash] mapping
     # @return [String, NilClass] full relator string
     def translate_relator(relator_code, mapping)
-      return unless relator_code.present?
+      return if relator_code.blank?
 
       mapping[relator_code.to_sym]
     end

data/pennmarc.gemspec

@@ -2,7 +2,7 @@
 
 Gem::Specification.new do |s|
   s.name        = 'pennmarc'
-  s.version     = '0.0.2'
+  s.version     = '1.0.1'
   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.'
@@ -19,5 +19,5 @@ Gem::Specification.new do |s|
   s.add_dependency 'marc', '~> 1.2'
   s.add_dependency 'nokogiri', '~> 1.15'
 
-  s.add_development_dependency 'rspec', '~> 3.12'
+  s.metadata['rubygems_mfa_required'] = 'true'
 end

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

@@ -17,11 +17,10 @@ describe 'PennMARC::Citation' do
   end
 
   describe '.cite_as_show' do
-    let(:record) { marc_record fields: [marc_field(tag: '524', subfields: {a: 'Perkins Historical Archive, Box 2'})] }
+    let(:record) { marc_record fields: [marc_field(tag: '524', subfields: { a: 'Perkins Historical Archive, Box 2' })] }
 
     it 'returns expected citation values' do
       expect(helper.cite_as_show(record)).to contain_exactly('Perkins Historical Archive, Box 2')
     end
   end
 end
-