use ActiveModel for Sword::Metadata::PersonalName and memoize expensive parses

- specs should not presumptively duplicate parser xpath data
- parse advisor names as per author names in ProQuest metadata parser

Author: barmintor

lib/sword/metadata/personal_name.rb

@@ -3,14 +3,16 @@
 module Sword
   module Metadata
     class PersonalName
-      attr_accessor :first_name,
-                    :full_name_naf_format,
-                    :last_name,
-                    :middle_name,
-                    :role,
-                    # type can be primary, alternate, etc. Kinda freeform
-                    :type,
-                    :uni
+      include ActiveModel::Model
+      include ActiveModel::Attributes
+      attribute :first_name, :string
+      attribute :full_name_naf_format, :string
+      attribute :last_name, :string
+      attribute :middle_name, :string
+      attribute :role, :string
+      # type can be primary, alternate, etc. Kinda freeform
+      attribute :type, :string
+      attribute :uni, :string
     end
   end
 end

lib/sword/mets/proquest_constants.rb

@@ -5,10 +5,12 @@ module Sword::Mets::ProquestConstants
   OTHER_MDTYPE = 'PROQUEST'
   XPATH_INFO = { namespace: { 'etdsword' => 'http://www.etdadmin.com/ns/etdsword' },
                  abstract: '//etdsword:DISS_abstract',
+                 advisor_name: '//etdsword:DISS_advisor/etdsword:DISS_name',
                  author_name: '//etdsword:DISS_author/etdsword:DISS_name',
                  first_name: './etdsword:DISS_fname',
                  middle_name: './etdsword:DISS_middle',
                  subjects: '//etdsword:DISS_cat_desc',
                  surname: './etdsword:DISS_surname',
-                 title: '//etdsword:DISS_title' }.freeze
+                 title: '//etdsword:DISS_title'
+               }.freeze
 end

lib/sword/mets/proquest_etd_xml_data_element.rb

@@ -1,36 +1,44 @@
 # frozen_string_literal: true
 
 class Sword::Mets::ProquestEtdXmlDataElement < Sword::Mets::XmlDataElement
-  attr_accessor :authors,
-                :subjects
+  include Sword::Mets::ProquestConstants
 
-  def initialize(nokogiri_xml_element, xpath_info)
+  def initialize(nokogiri_xml_element, xpath_info = Sword::Mets::ProquestConstants::XPATH_INFO)
     super
-    @authors = []
-    @subjects = []
+    @authors = nil
+    @subjects = nil
   end
 
-  def parse_authors
-    diss_authors = @xml_data.xpath(@xpath_info[:author_name], @xpath_info[:namespace])
-    diss_authors.each do |author|
-      person = create_personal_author(author)
-      person.role = 'author'
-      @authors << person
+  # Override
+  def authors
+    @authors ||= begin
+      diss_authors = @xml_data.xpath(@xpath_info[:author_name], @xpath_info[:namespace])
+      diss_authors.map { |author| create_personal_name(author, role: 'author') }
     end
   end
 
-  def create_personal_author(author)
-    person = Sword::Metadata::PersonalName.new
-    person.first_name = author.xpath(@xpath_info[:first_name], @xpath_info[:namespace]).text
-    person.middle_name = author.xpath(@xpath_info[:middle_name], @xpath_info[:namespace]).text
-    person.last_name = author.xpath(@xpath_info[:surname], @xpath_info[:namespace]).text
-    person
+  def advisors
+    @advisors ||= begin
+      diss_advisors = @xml_data.xpath(@xpath_info[:advisor_name], @xpath_info[:namespace])
+      diss_advisors.map { |author| create_personal_name(author, role: 'advisor') }
+    end
   end
 
-  def parse_subjects
-    @diss_cat_descs = @xml_data.xpath(@xpath_info[:subjects], @xpath_info[:namespace])
-    @diss_cat_descs.each do |diss_cat_desc|
-      @subjects.push diss_cat_desc.text
+  # Override
+  def subjects
+    @subjects ||= begin
+      @diss_cat_descs = @xml_data.xpath(@xpath_info[:subjects], @xpath_info[:namespace])
+      @diss_cat_descs.map(&:text)
     end
   end
+
+  private
+  def create_personal_name(author, **other_attrs)
+    author_attrs = other_attrs.merge({
+      first_name: author.xpath(@xpath_info[:first_name], @xpath_info[:namespace]).text,
+      middle_name: author.xpath(@xpath_info[:middle_name], @xpath_info[:namespace]).text,
+      last_name: author.xpath(@xpath_info[:surname], @xpath_info[:namespace]).text,
+    })
+    Sword::Metadata::PersonalName.new(**author_attrs)
+  end
 end

lib/sword/mets/xml_data_element.rb

@@ -5,7 +5,14 @@ class Sword::Mets::XmlDataElement
                 :advisors,
                 :authors,
                 :subjects,
-                :title
+                :title,
+                :date_conferred,
+                :degree,
+                :embargo_code,
+                :embargo_release_date,
+                :institution_department_name,
+                :institution_name,
+                :institution_school_code
 
   def initialize(nokogiri_xml_element, xpath_info)
     @xml_data = nokogiri_xml_element

spec/sword/mets/proquest_etd_xml_data_element_spec.rb

@@ -3,56 +3,39 @@
 require 'rails_helper'
 
 RSpec.describe Sword::Mets::ProquestEtdXmlDataElement do
-  let(:xpath_info) do
-    { namespace: { 'etdsword' => 'http://www.etdadmin.com/ns/etdsword' },
-      abstract: '//etdsword:DISS_abstract',
-      author_name: '//etdsword:DISS_author/etdsword:DISS_name',
-      first_name: 'etdsword:DISS_fname',
-      middle_name: 'etdsword:DISS_middle',
-      subjects: '//etdsword:DISS_cat_desc',
-      surname: 'etdsword:DISS_surname',
-      title: '//etdsword:DISS_title' }
+  let(:pq_xml_data) { described_class.new(@xml_data_proquest, Sword::Mets::ProquestConstants::XPATH_INFO) }
+
+  let(:author_attrs) { { first_name: 'Ariana', middle_name: 'Cecilia', last_name: 'Gavin', role: 'author' } }
+  let(:expected_first_author) { Sword::Metadata::PersonalName.new(author_attrs) }
+
+  let(:advisor_attrs) { { first_name: 'Henry', middle_name: 'M', last_name: 'Colecraft', role: 'advisor' } }
+  let(:expected_first_advisor) { Sword::Metadata::PersonalName.new(advisor_attrs) }
+
+  before(:context) do
+    @pq_mets_file = Sword::Mets::MetsFile.new(file_fixture('xml/mets/PQ_mets.xml').read)
+    # @xml_data_proquest = @pq_mets_file.find_md_wrap_xml_data_elements(mdtype: 'OTHER', other_mdtype: 'PROQUEST')
+    @xml_data_proquest = @pq_mets_file.find_md_wrap_xml_data_element(mdtype: 'OTHER', other_mdtype: 'PROQUEST')
+  end
+
+  it 'sets the abstract correctly' do
+    expect(pq_xml_data.abstract).to include('relief of Rad inhibition of cardiac')
+  end
+
+  it 'sets the title correctly' do
+    expect(pq_xml_data.title).to include('Unraveling the logic')
+  end
+
+  it 'sets the subjects correctly' do
+    expect(pq_xml_data.instance_variable_get(:@subjects)).to be_nil
+    expect(pq_xml_data.subjects).to contain_exactly('Molecular biology', 'Pharmacology', 'Physiology')
   end
 
-  describe 'at initialization' do
-    before(:context) do
-      @pq_mets_file = Sword::Mets::MetsFile.new(file_fixture('xml/mets/PQ_mets.xml').read)
-      # @xml_data_proquest = @pq_mets_file.find_md_wrap_xml_data_elements(mdtype: 'OTHER', other_mdtype: 'PROQUEST')
-      @xml_data_proquest = @pq_mets_file.find_md_wrap_xml_data_element(mdtype: 'OTHER', other_mdtype: 'PROQUEST')
-    end
-
-    it 'sets the abstract correctly' do
-      pq_xml_data = described_class.new(@xml_data_proquest, xpath_info)
-      expect(pq_xml_data.abstract).to include('relief of Rad inhibition of cardiac')
-    end
-
-    it 'sets the title correctly' do
-      pq_xml_data = described_class.new(@xml_data_proquest, xpath_info)
-      expect(pq_xml_data.title).to include('Unraveling the logic')
-    end
+  it 'sets the authors correctly' do
+    expect(pq_xml_data.instance_variable_get(:@authors)).to be_nil
+    expect(pq_xml_data.authors.first.attributes).to eql(expected_first_author.attributes)
   end
 
-  describe 'after parsing' do
-    before(:context) do
-      @pq_mets_file = Sword::Mets::MetsFile.new(file_fixture('xml/mets/PQ_mets.xml').read)
-      # @xml_data_proquest = @pq_mets_file.find_md_wrap_xml_data_elements(mdtype: 'OTHER', other_mdtype: 'PROQUEST')
-      @xml_data_proquest = @pq_mets_file.find_md_wrap_xml_data_element(mdtype: 'OTHER', other_mdtype: 'PROQUEST')
-    end
-
-    it 'sets the subjects correctly' do
-      pq_xml_data = described_class.new(@xml_data_proquest, xpath_info)
-      pq_xml_data.parse_subjects
-      expect(pq_xml_data.subjects).to include('Molecular biology')
-      expect(pq_xml_data.subjects).to include('Pharmacology')
-      expect(pq_xml_data.subjects).to include('Physiology')
-    end
-
-    it 'sets the authors correctly' do
-      pq_xml_data = described_class.new(@xml_data_proquest, xpath_info)
-      pq_xml_data.parse_authors
-      expect(pq_xml_data.authors.first.first_name).to include('Ariana')
-      expect(pq_xml_data.authors.first.middle_name).to include('Cecilia')
-      expect(pq_xml_data.authors.first.last_name).to include('Gavin')
-    end
+  it 'sets the advisors correctly' do
+    expect(pq_xml_data.advisors.first.attributes).to eql(expected_first_advisor.attributes)
   end
 end

spec/sword/mets/xml_data_element_spec.rb

@@ -9,21 +9,14 @@
       # @xml_data_proquest = @pq_mets_file.find_md_wrap_xml_data_elements(mdtype: 'OTHER', other_mdtype: 'PROQUEST')
       @xml_data_proquest = @pq_mets_file.find_md_wrap_xml_data_element(mdtype: 'OTHER', other_mdtype: 'PROQUEST')
     end
+    let(:pq_xml_data) { described_class.new(@xml_data_proquest, Sword::Mets::ProquestConstants::XPATH_INFO) }
 
     it 'sets the abstract correctly' do
-      xpath_info = { namespace: { 'etdsword' => 'http://www.etdadmin.com/ns/etdsword' },
-                     abstract: '//etdsword:DISS_abstract' }
-      pq_xml_data = described_class.new(@xml_data_proquest, xpath_info)
       expect(pq_xml_data.abstract).to include('relief of Rad inhibition of cardiac')
-      # expect(true).to be true
     end
 
     it 'sets the title correctly' do
-      xpath_info = { namespace: { 'etdsword' => 'http://www.etdadmin.com/ns/etdsword' },
-                     title: '//etdsword:DISS_title' }
-      pq_xml_data = described_class.new(@xml_data_proquest, xpath_info)
       expect(pq_xml_data.title).to include('Unraveling the logic')
-      # expect(true).to be true
     end
   end
 end