gdor-indexer 0.3.0 → 0.4.0

data/spec/unit/mods_pub_fields_spec.rb

@@ -0,0 +1,257 @@
+require 'spec_helper'
+
+describe GDor::Indexer::ModsFields do
+  let(:fake_druid) { 'oo000oo0000' }
+  let(:ns_decl) { "xmlns='#{Mods::MODS_NS}'" }
+  let(:mods_xml) { "<mods #{ns_decl}><note>gdor_mods_fields testing</note></mods>" }
+  let(:mods_origin_info_start_str) { "<mods #{ns_decl}><originInfo>" }
+  let(:mods_origin_info_end_str) { '</originInfo></mods>' }
+
+  def sdb_for_mods(m)
+    resource = Harvestdor::Indexer::Resource.new(double, fake_druid)
+    allow(resource).to receive(:public_xml).and_return(nil)
+    allow(resource).to receive(:mods).and_return(Nokogiri::XML(m))
+    i = Harvestdor::Indexer.new
+    i.logger.level = Logger::WARN
+    allow(resource).to receive(:indexer).and_return(i)
+    lgr = Logger.new(StringIO.new)
+    lgr.level = Logger::WARN
+    GDor::Indexer::SolrDocBuilder.new(resource, lgr)
+  end
+
+  let(:sdb) { sdb_for_mods(mods_xml) }
+
+  context 'publication date fields' do
+
+    RSpec.shared_examples 'expected' do |solr_field_sym, mods_field_val, exp_val|
+      it "#{exp_val} for #{mods_field_val}" do
+        m = mods_origin_info_start_str +
+              "<dateIssued>#{mods_field_val}</dateIssued>" +
+            mods_origin_info_end_str
+        sdb = sdb_for_mods(m)
+        expect(sdb.doc_hash_from_mods[solr_field_sym]).to eq exp_val
+      end
+    end
+
+    context 'pub_date_sort' do
+      it 'calls Stanford::Mods::Record instance pub_date_sortable_string(false)' do
+        expect(sdb.smods_rec).to receive(:pub_date_sortable_string).with(false)
+        sdb.doc_hash_from_mods[:pub_date_sort]
+      end
+      it 'includes approx dates' do
+        m = mods_origin_info_start_str +
+              "<dateIssued qualifier='approximate'>1945</dateIssued>" +
+            mods_origin_info_end_str
+        sdb = sdb_for_mods(m)
+        expect(sdb.doc_hash_from_mods[:pub_date_sort]).to eq('1945')
+      end
+      it 'takes single dateCreated' do
+        m = mods_origin_info_start_str +
+              "<dateCreated>1904</dateCreated>" +
+            mods_origin_info_end_str
+        sdb = sdb_for_mods(m)
+        expect(sdb.doc_hash_from_mods[:pub_date_sort]).to eq('1904')
+      end
+      it_behaves_like 'expected', :pub_date_sort, '1945', '1945'
+      it_behaves_like 'expected', :pub_date_sort, '945', '0945'
+      it_behaves_like 'expected', :pub_date_sort, '66', '0066'
+      it_behaves_like 'expected', :pub_date_sort, '5', '0005'
+      it_behaves_like 'expected', :pub_date_sort, '0', '0000'
+      # these negative values are for String lexical sorting as this is a string
+      it_behaves_like 'expected', :pub_date_sort, '-4', '-996'
+      it_behaves_like 'expected', :pub_date_sort, '-15', '-985'
+      it_behaves_like 'expected', :pub_date_sort, '-666', '-334'
+      it_behaves_like 'expected', :pub_date_sort, '16--', '1600'
+      it_behaves_like 'expected', :pub_date_sort, '9--', '0900'
+      it_behaves_like 'expected', :pub_date_sort, '19th century', '1800'
+      it_behaves_like 'expected', :pub_date_sort, '9th century', '0800'
+      # -(1000 - |yyy|) for BC dates
+      it_behaves_like 'expected', :pub_date_sort, '300 B.C.', '-700'
+    end
+
+    context 'single valued pub year facets' do
+      let(:mods) do
+        mods_origin_info_start_str +
+          "<dateIssued qualifier=\"approximate\">1500</dateIssued>
+          <dateIssued>2000</dateIssued>" +
+        mods_origin_info_end_str
+      end
+      it 'pub_year_no_approx_isi calls Stanford::Mods::Record instance pub_date_facet_single_value(true)' do
+        sdb = sdb_for_mods(mods)
+        expect(sdb.smods_rec).to receive(:pub_date_facet_single_value).with(true).and_call_original
+        allow(sdb.smods_rec).to receive(:pub_date_facet_single_value).with(false) # for other flavor
+        expect(sdb.doc_hash_from_mods[:pub_year_no_approx_isi]).to eq '2000'
+      end
+      it 'pub_year_w_approx_isi calls Stanford::Mods::Record instance pub_date_facet_single_value(false)' do
+        sdb = sdb_for_mods(mods)
+        expect(sdb.smods_rec).to receive(:pub_date_facet_single_value).with(false).and_call_original
+        allow(sdb.smods_rec).to receive(:pub_date_facet_single_value).with(true) # for other flavor
+        expect(sdb.doc_hash_from_mods[:pub_year_w_approx_isi]).to eq '1500'
+      end
+      RSpec.shared_examples "single pub year facet" do |field_sym|
+        it_behaves_like 'expected', field_sym, '1945', '1945'
+        it_behaves_like 'expected', field_sym, '945', '945'
+        it_behaves_like 'expected', field_sym, '66', '66'
+        it_behaves_like 'expected', field_sym, '5', '5'
+        it_behaves_like 'expected', field_sym, '0', '0'
+        it_behaves_like 'expected', field_sym, '-4', '4 B.C.'
+        it_behaves_like 'expected', field_sym, '-15', '15 B.C.'
+        it_behaves_like 'expected', field_sym, '-666', '666 B.C.'
+        it_behaves_like 'expected', field_sym, '16--', '17th century'
+        it_behaves_like 'expected', field_sym, '8--', '9th century'
+        it_behaves_like 'expected', field_sym, '19th century', '19th century'
+        it_behaves_like 'expected', field_sym, '9th century', '9th century'
+        it_behaves_like 'expected', field_sym, '300 B.C.', '300 B.C.'
+      end
+      it_behaves_like "single pub year facet", :pub_year_no_approx_isi
+      it_behaves_like "single pub year facet", :pub_year_w_approx_isi
+    end
+
+    context 'pub_year_tisim for date slider' do
+      it 'should handle multiple pub dates (to be implemented - esp for date slider)'
+
+      # FIXME:  it should be using a method approp for date slider values, not single value
+      it 'pub_year_tisim calls Stanford::Mods::Record instance pub_date_sortable_string(false)' do
+        expect(sdb.smods_rec).to receive(:pub_date_sortable_string).with(false)
+        sdb.doc_hash_from_mods[:pub_year_tisim]
+      end
+      it 'includes approx dates' do
+        m = mods_origin_info_start_str +
+              "<dateIssued qualifier='approximate'>1945</dateIssued>" +
+          mods_origin_info_end_str
+        sdb = sdb_for_mods(m)
+        expect(sdb.doc_hash_from_mods[:pub_year_tisim]).to eq('1945')
+      end
+      it 'takes single dateCreated' do
+        m = mods_origin_info_start_str +
+              "<dateCreated>1904</dateCreated>" +
+            mods_origin_info_end_str
+        sdb = sdb_for_mods(m)
+        expect(sdb.doc_hash_from_mods[:pub_year_tisim]).to eq('1904')
+      end
+      it 'ignores B.C. dates' do
+        m = mods_origin_info_start_str +
+              "<dateCreated>300 B.C.</dateCreated>" +
+          mods_origin_info_end_str
+        sdb = sdb_for_mods(m)
+        expect(sdb.doc_hash_from_mods).not_to have_key(:pub_year_tisim)
+        expect(sdb.doc_hash_from_mods[:pub_year_tisim]).to be_nil
+      end
+      it_behaves_like 'expected', :pub_year_tisim, '1945', '1945'
+      it_behaves_like 'expected', :pub_year_tisim, '945', '0945'
+      it_behaves_like 'expected', :pub_year_tisim, '66', '0066'
+      it_behaves_like 'expected', :pub_year_tisim, '5', '0005'
+      it_behaves_like 'expected', :pub_year_tisim, '0', '0000'
+      it_behaves_like 'expected', :pub_year_tisim, '-4', nil
+      it_behaves_like 'expected', :pub_year_tisim, '-15', nil
+      it_behaves_like 'expected', :pub_year_tisim, '-666', nil
+      it_behaves_like 'expected', :pub_year_tisim, '16--', '1600'
+      it_behaves_like 'expected', :pub_year_tisim, '9--', '0900'
+      it_behaves_like 'expected', :pub_year_tisim, '19th century', '1800'
+      it_behaves_like 'expected', :pub_year_tisim, '9th century', '0800'
+      it_behaves_like 'expected', :pub_year_tisim, 'Text dated June 4, 1594; miniatures added by 1596', '1594'
+      it_behaves_like 'expected', :pub_year_tisim, 'Aug. 3rd, 1886', '1886'
+      it_behaves_like 'expected', :pub_year_tisim, 'Aug. 3rd, [18]86?', '1886'
+      it_behaves_like 'expected', :pub_year_tisim, 'early 1890s', '1890'
+      it_behaves_like 'expected', :pub_year_tisim, '1865-6', '1865' # FIXME:  should be both years
+    end
+
+    context 'creation_year_isi' do
+      it 'creation_year_isi calls Stanford::Mods::Record pub_date_best_sort_str_value for dateCreated elements' do
+        m = mods_origin_info_start_str +
+              "<dateCreated qualifier='approximate'>1500</dateCreated>
+              <dateIssued qualifier='approximate'>2000</dateIssued>" +
+            mods_origin_info_end_str
+        sdb = sdb_for_mods(m)
+        expect(sdb.smods_rec).to receive(:pub_date_best_sort_str_value).at_least(2).times.and_call_original
+        expect(sdb.doc_hash_from_mods[:creation_year_isi]).to eq '1500'
+      end
+      RSpec.shared_examples 'expected for dateCreated' do |mods_field_val, exp_val|
+        it "#{exp_val} for #{mods_field_val}" do
+          m = mods_origin_info_start_str +
+                "<dateCreated>#{mods_field_val}</dateCreated>" +
+              mods_origin_info_end_str
+          sdb = sdb_for_mods(m)
+          expect(sdb.doc_hash_from_mods[:creation_year_isi]).to eq exp_val
+        end
+      end
+      it_behaves_like 'expected for dateCreated', '1945', '1945'
+      # note that it removes leading zeros
+      it_behaves_like 'expected for dateCreated', '945', '945'
+      it_behaves_like 'expected for dateCreated', '66', '66'
+      it_behaves_like 'expected for dateCreated', '5', '5'
+      it_behaves_like 'expected for dateCreated', '0', '0'
+      it_behaves_like 'expected for dateCreated', '-4', '-4'
+      it_behaves_like 'expected for dateCreated', '-15', '-15'
+      it_behaves_like 'expected for dateCreated', '-666', '-666'
+      it_behaves_like 'expected for dateCreated', '16--', '1600'
+      it_behaves_like 'expected for dateCreated', '9--', '900'
+      it_behaves_like 'expected for dateCreated', '19th century', '1800'
+      it_behaves_like 'expected for dateCreated', '9th century', '800'
+      it_behaves_like 'expected for dateCreated', 'blah June 4, 1594; blah 1596', '1594'
+      it_behaves_like 'expected for dateCreated', 'Aug. 3rd, 1886', '1886'
+      it_behaves_like 'expected for dateCreated', 'Aug. 3rd, [18]86?', '1886'
+      it_behaves_like 'expected for dateCreated', 'early 1890s', '1890'
+      it_behaves_like 'expected for dateCreated', '1865-6', '1865'
+      # note: B.C. becomes a regular negative number
+      it_behaves_like 'expected for dateCreated', '300 B.C.', '-300'
+    end
+
+    context 'publication_year_isi' do
+      it 'publication_year_isi calls Stanford::Mods::Record pub_date_best_sort_str_value for dateIssued elements' do
+        m = mods_origin_info_start_str +
+              "<dateCreated qualifier='approximate'>1500</dateCreated>
+              <dateIssued qualifier='approximate'>2000</dateIssued>" +
+            mods_origin_info_end_str
+        sdb = sdb_for_mods(m)
+        expect(sdb.smods_rec).to receive(:pub_date_best_sort_str_value).at_least(2).times.and_call_original
+        expect(sdb.doc_hash_from_mods[:publication_year_isi]).to eq '2000'
+      end
+      it_behaves_like 'expected', :publication_year_isi, '1945', '1945'
+      # note that it removes leading zeros
+      it_behaves_like 'expected', :publication_year_isi, '945', '945'
+      it_behaves_like 'expected', :publication_year_isi, '66', '66'
+      it_behaves_like 'expected', :publication_year_isi, '5', '5'
+      it_behaves_like 'expected', :publication_year_isi, '0', '0'
+      it_behaves_like 'expected', :publication_year_isi, '-4', '-4'
+      it_behaves_like 'expected', :publication_year_isi, '-15', '-15'
+      it_behaves_like 'expected', :publication_year_isi, '-666', '-666'
+      it_behaves_like 'expected', :publication_year_isi, '16--', '1600'
+      it_behaves_like 'expected', :publication_year_isi, '9--', '900'
+      it_behaves_like 'expected', :publication_year_isi, '19th century', '1800'
+      it_behaves_like 'expected', :publication_year_isi, '9th century', '800'
+      it_behaves_like 'expected', :publication_year_isi, 'blah June 4, 1594; blah 1596', '1594'
+      it_behaves_like 'expected', :publication_year_isi, 'Aug. 3rd, 1886', '1886'
+      it_behaves_like 'expected', :publication_year_isi, 'Aug. 3rd, [18]86?', '1886'
+      it_behaves_like 'expected', :publication_year_isi, 'early 1890s', '1890'
+      it_behaves_like 'expected', :publication_year_isi, '1865-6', '1865'
+      # note: B.C. becomes a regular negative number
+      it_behaves_like 'expected', :publication_year_isi, '300 B.C.', '-300'
+    end
+  end # publication date fields
+
+  context 'imprint_display' do
+    # FIXME:  it should be using a method returning a better string than just year
+    it 'imprint_display calls deprecated Stanford::Mods::Record instance pub_date_display' do
+      expect(sdb.smods_rec).to receive(:pub_date_display)
+      sdb.doc_hash_from_mods[:imprint_display]
+    end
+    it_behaves_like 'expected', :imprint_display, '1945', '1945'
+    it_behaves_like 'expected', :imprint_display, '945', '945'
+    it_behaves_like 'expected', :imprint_display, '66', '66'
+    it_behaves_like 'expected', :imprint_display, '5', '5'
+    it_behaves_like 'expected', :imprint_display, '0', '0'
+    it_behaves_like 'expected', :imprint_display, '-4', '-4'
+    it_behaves_like 'expected', :imprint_display, '-15', '-15'
+    it_behaves_like 'expected', :imprint_display, '-666', '-666'
+    it_behaves_like 'expected', :imprint_display, '16--', '16--'
+    it_behaves_like 'expected', :imprint_display, '9--', '9--'
+    it_behaves_like 'expected', :imprint_display, '19th century', '19th century'
+    it_behaves_like 'expected', :imprint_display, '9th century', '9th century'
+    it_behaves_like 'expected', :imprint_display, 'blah June 4, 1594; blah 1596', 'blah June 4, 1594; blah 1596'
+    it_behaves_like 'expected', :imprint_display, 'Aug. 3rd, 1886', 'Aug. 3rd, 1886'
+    it_behaves_like 'expected', :imprint_display, 'Aug. 3rd, [18]86?', 'Aug. 3rd, [18]86?'
+    it_behaves_like 'expected', :imprint_display, 'early 1890s', 'early 1890s'
+    it_behaves_like 'expected', :imprint_display, '1865-6', '1865-6'
+  end
+end

data/spec/unit/mods_subject_fields_spec.rb

@@ -0,0 +1,279 @@
+require 'spec_helper'
+
+describe GDor::Indexer::ModsFields do
+  let(:fake_druid) { 'oo000oo0000' }
+  let(:ns_decl) { "xmlns='#{Mods::MODS_NS}'" }
+  let(:mods_xml) { "<mods #{ns_decl}><note>gdor_mods_fields testing</note></mods>" }
+
+  def sdb_for_mods(m)
+    resource = Harvestdor::Indexer::Resource.new(double, fake_druid)
+    allow(resource).to receive(:public_xml).and_return(nil)
+    allow(resource).to receive(:mods).and_return(Nokogiri::XML(m))
+    i = Harvestdor::Indexer.new
+    i.logger.level = Logger::WARN
+    allow(resource).to receive(:indexer).and_return(i)
+    lgr = Logger.new(StringIO.new)
+    lgr.level = Logger::WARN
+    GDor::Indexer::SolrDocBuilder.new(resource, lgr)
+  end
+
+  context 'subject fields' do
+    let(:genre) { 'genre top level' }
+    let(:cart_coord) { '6 00 S, 71 30 E' }
+    let(:s_genre) { 'genre in subject' }
+    let(:geo) { 'Somewhere' }
+    let(:geo_code) { 'us' }
+    let(:hier_geo_country) { 'France' }
+    let(:s_name) { 'name in subject' }
+    let(:occupation) { 'worker bee' }
+    let(:temporal) { 'temporal' }
+    let(:s_title) { 'title in subject' }
+    let(:topic) { 'topic' }
+    let(:m) do
+      "<mods #{ns_decl}>
+        <genre>#{genre}</genre>
+        <subject><cartographics><coordinates>#{cart_coord}</coordinates></cartographics></subject>
+        <subject><genre>#{s_genre}</genre></subject>
+        <subject><geographic>#{geo}</geographic></subject>
+        <subject><geographicCode authority='iso3166'>#{geo_code}</geographicCode></subject>
+        <subject><hierarchicalGeographic><country>#{hier_geo_country}</country></hierarchicalGeographic></subject>
+        <subject><name><namePart>#{s_name}</namePart></name></subject>
+        <subject><occupation>#{occupation}</occupation></subject>
+        <subject><temporal>#{temporal}</temporal></subject>
+        <subject><titleInfo><title>#{s_title}</title></titleInfo></subject>
+        <subject><topic>#{topic}</topic></subject>
+        <typeOfResource>still image</typeOfResource>
+      </mods>"
+    end
+    let(:m_no_subject) { "<mods #{ns_decl}><note>notit</note></mods>" }
+    let(:sdb) { sdb_for_mods(m) }
+    let(:subject_doc_hash) { sdb.doc_hash_from_mods }
+
+    it 'calls the appropriate methods in stanford-mods to populate the Solr fields' do
+      expect(sdb.smods_rec).to receive(:topic_search)
+      expect(sdb.smods_rec).to receive(:geographic_search)
+      expect(sdb.smods_rec).to receive(:subject_other_search)
+      expect(sdb.smods_rec).to receive(:subject_other_subvy_search)
+      expect(sdb.smods_rec).to receive(:subject_all_search)
+      expect(sdb.smods_rec).to receive(:topic_facet)
+      expect(sdb.smods_rec).to receive(:geographic_facet)
+      expect(sdb.smods_rec).to receive(:era_facet)
+      sdb.doc_hash_from_mods
+    end
+    context 'search fields' do
+      context 'topic_search' do
+        it 'onlies include genre and topic' do
+          expect(subject_doc_hash[:topic_search]).to match_array [genre, topic]
+        end
+        context 'functional tests checking results from stanford-mods methods' do
+          it 'is nil if there are no values in the MODS' do
+            sdb = sdb_for_mods(m_no_subject)
+            expect(sdb.doc_hash_from_mods[:topic_search]).to be_nil
+          end
+          it 'does not be nil if there are only subject/topic elements (no <genre>)' do
+            m = "<mods #{ns_decl}><subject><topic>#{topic}</topic></subject></mods>"
+            sdb = sdb_for_mods(m)
+            expect(sdb.doc_hash_from_mods[:topic_search]).to match_array [topic]
+          end
+          it 'does not be nil if there are only <genre> elements (no subject/topic elements)' do
+            m = "<mods #{ns_decl}><genre>#{genre}</genre></mods>"
+            sdb = sdb_for_mods(m)
+            expect(sdb.doc_hash_from_mods[:topic_search]).to match_array [genre]
+          end
+          it 'has a separate value for each topic subelement' do
+            m = "<mods #{ns_decl}>
+            <subject>
+              <topic>first</topic>
+              <topic>second</topic>
+            </subject>
+            <subject><topic>third</topic></subject>
+            </mods>"
+            sdb = sdb_for_mods(m)
+            expect(sdb.doc_hash_from_mods[:topic_search]).to match_array %w(first second third)
+          end
+        end # functional tests checking results from stanford-mods methods
+      end # topic_search
+
+      context 'geographic_search' do
+        it 'includes geographic and hierarchicalGeographic' do
+          expect(subject_doc_hash[:geographic_search]).to match_array [geo, hier_geo_country]
+        end
+        it 'calls sw_geographic_search (from stanford-mods gem)' do
+          m = "<mods #{ns_decl}><subject><geographic>#{geo}</geographic></subject></mods>"
+          sdb = sdb_for_mods(m)
+          expect(sdb.smods_rec).to receive(:sw_geographic_search).at_least(1).times
+          sdb.doc_hash_from_mods
+        end
+        it "logs an info message when it encounters a geographicCode encoding it doesn't translate" do
+          m = "<mods #{ns_decl}><subject><geographicCode authority='iso3166'>ca</geographicCode></subject></mods>"
+          sdb = sdb_for_mods(m)
+          expect(sdb.smods_rec.sw_logger).to receive(:info).with(/#{fake_druid} has subject geographicCode element with untranslated encoding \(iso3166\): <geographicCode authority=.*>ca<\/geographicCode>/).at_least(1).times
+          sdb.doc_hash_from_mods
+        end
+      end # geographic_search
+
+      context 'subject_other_search' do
+        it 'includes occupation, subject names, and subject titles' do
+          expect(subject_doc_hash[:subject_other_search]).to match_array [occupation, s_name, s_title]
+        end
+        context 'functional tests checking results from stanford-mods methods' do
+          it 'is nil if there are no values in the MODS' do
+            sdb = sdb_for_mods(mods_xml)
+            expect(sdb.doc_hash_from_mods[:subject_other_search]).to be_nil
+          end
+          it 'does not be nil if there are only subject/name elements' do
+            m = "<mods #{ns_decl}><subject><name><namePart>#{s_name}</namePart></name></subject></mods>"
+            sdb = sdb_for_mods(m)
+            expect(sdb.doc_hash_from_mods[:subject_other_search]).to match_array [s_name]
+          end
+          it 'does not be nil if there are only subject/occupation elements' do
+            m = "<mods #{ns_decl}><subject><occupation>#{occupation}</occupation></subject></mods>"
+            sdb = sdb_for_mods(m)
+            expect(sdb.doc_hash_from_mods[:subject_other_search]).to match_array [occupation]
+          end
+          it 'does not be nil if there are only subject/titleInfo elements' do
+            m = "<mods #{ns_decl}><subject><titleInfo><title>#{s_title}</title></titleInfo></subject></mods>"
+            sdb = sdb_for_mods(m)
+            expect(sdb.doc_hash_from_mods[:subject_other_search]).to match_array [s_title]
+          end
+          it 'has a separate value for each occupation subelement' do
+            m = "<mods #{ns_decl}>
+            <subject>
+              <occupation>first</occupation>
+              <occupation>second</occupation>
+            </subject>
+            <subject><occupation>third</occupation></subject>
+            </mods>"
+            sdb = sdb_for_mods(m)
+            expect(sdb.doc_hash_from_mods[:subject_other_search]).to match_array %w(first second third)
+          end
+        end # functional tests checking results from stanford-mods methods
+      end # subject_other_search
+
+      context 'subject_other_subvy_search' do
+        it 'includes temporal and genre SUBelement' do
+          expect(subject_doc_hash[:subject_other_subvy_search]).to match_array [temporal, s_genre]
+        end
+        context 'functional tests checking results from stanford-mods methods' do
+          it 'is nil if there are no values in the MODS' do
+            sdb = sdb_for_mods(mods_xml)
+            expect(sdb.doc_hash_from_mods[:subject_other_subvy_search]).to be_nil
+          end
+          it 'does not be nil if there are only subject/temporal elements (no subject/genre)' do
+            m = "<mods #{ns_decl}><subject><temporal>#{temporal}</temporal></subject></mods>"
+            sdb = sdb_for_mods(m)
+            expect(sdb.doc_hash_from_mods[:subject_other_subvy_search]).to match_array [temporal]
+          end
+          it 'does not be nil if there are only subject/genre elements (no subject/temporal)' do
+            m = "<mods #{ns_decl}><subject><genre>#{s_genre}</genre></subject></mods>"
+            sdb = sdb_for_mods(m)
+            expect(sdb.doc_hash_from_mods[:subject_other_subvy_search]).to match_array [s_genre]
+          end
+          context 'genre subelement' do
+            it 'has a separate value for each genre element' do
+              m = "<mods #{ns_decl}>
+              <subject>
+                <genre>first</genre>
+                <genre>second</genre>
+              </subject>
+              <subject><genre>third</genre></subject>
+              </mods>"
+              sdb = sdb_for_mods(m)
+              expect(sdb.doc_hash_from_mods[:subject_other_subvy_search]).to match_array %w(first second third)
+            end
+          end # genre subelement
+        end # "functional tests checking results from stanford-mods methods"
+      end # subject_other_subvy_search
+
+      context 'subject_all_search' do
+        it 'contains top level <genre> element data' do
+          expect(subject_doc_hash[:subject_all_search]).to include(genre)
+        end
+        it 'does not contain cartographic sub element' do
+          expect(subject_doc_hash[:subject_all_search]).not_to include(cart_coord)
+        end
+        it 'does not include codes from hierarchicalGeographic sub element' do
+          expect(subject_doc_hash[:subject_all_search]).not_to include(geo_code)
+        end
+        it 'contains all other subject subelement data' do
+          expect(subject_doc_hash[:subject_all_search]).to include(s_genre)
+          expect(subject_doc_hash[:subject_all_search]).to include(geo)
+          expect(subject_doc_hash[:subject_all_search]).to include(hier_geo_country)
+          expect(subject_doc_hash[:subject_all_search]).to include(s_name)
+          expect(subject_doc_hash[:subject_all_search]).to include(occupation)
+          expect(subject_doc_hash[:subject_all_search]).to include(temporal)
+          expect(subject_doc_hash[:subject_all_search]).to include(s_title)
+          expect(subject_doc_hash[:subject_all_search]).to include(topic)
+        end
+      end # subject_all_search
+    end # search fields
+
+    context 'facet fields' do
+      context 'topic_facet' do
+        it 'includes topic subelement' do
+          expect(subject_doc_hash[:topic_facet]).to include(topic)
+        end
+        it 'includes sw_subject_names' do
+          expect(subject_doc_hash[:topic_facet]).to include(s_name)
+        end
+        it 'includes sw_subject_titles' do
+          expect(subject_doc_hash[:topic_facet]).to include(s_title)
+        end
+        it 'includes occupation subelement' do
+          expect(subject_doc_hash[:topic_facet]).to include(occupation)
+        end
+        it 'has the trailing punctuation removed' do
+          m = "<mods #{ns_decl}><subject>
+          <topic>comma,</topic>
+          <occupation>semicolon;</occupation>
+          <titleInfo><title>backslash \\</title></titleInfo>
+          <name><namePart>internal, punct;uation</namePart></name>
+          </subject></mods>"
+          sdb = sdb_for_mods(m)
+          doc_hash = sdb.doc_hash_from_mods
+          expect(doc_hash[:topic_facet]).to include('comma')
+          expect(doc_hash[:topic_facet]).to include('semicolon')
+          expect(doc_hash[:topic_facet]).to include('backslash')
+          expect(doc_hash[:topic_facet]).to include('internal, punct;uation')
+        end
+      end # topic_facet
+
+      context 'geographic_facet' do
+        it 'includes geographic subelement' do
+          expect(subject_doc_hash[:geographic_facet]).to include(geo)
+        end
+        it 'is like geographic_search with the trailing punctuation (and preceding spaces) removed' do
+          m = "<mods #{ns_decl}><subject>
+          <geographic>comma,</geographic>
+          <geographic>semicolon;</geographic>
+          <geographic>backslash \\</geographic>
+          <geographic>internal, punct;uation</geographic>
+          </subject></mods>"
+          sdb = sdb_for_mods(m)
+          doc_hash = sdb.doc_hash_from_mods
+          expect(doc_hash[:geographic_facet]).to include('comma')
+          expect(doc_hash[:geographic_facet]).to include('semicolon')
+          expect(doc_hash[:geographic_facet]).to include('backslash')
+          expect(doc_hash[:geographic_facet]).to include('internal, punct;uation')
+        end
+      end
+
+      it 'era_facet should be temporal subelement with the trailing punctuation removed' do
+        m = "<mods #{ns_decl}><subject>
+        <temporal>comma,</temporal>
+        <temporal>semicolon;</temporal>
+        <temporal>backslash \\</temporal>
+        <temporal>internal, punct;uation</temporal>
+        </subject></mods>"
+        sdb = sdb_for_mods(m)
+        doc_hash = sdb.doc_hash_from_mods
+        expect(doc_hash[:era_facet]).to include('comma')
+        expect(doc_hash[:era_facet]).to include('semicolon')
+        expect(doc_hash[:era_facet]).to include('backslash')
+        expect(doc_hash[:era_facet]).to include('internal, punct;uation')
+      end
+    end # facet fields
+  end # subject fields
+
+  # publication fields moved to mods_pub_fields_spec.rb
+end