Merge pull request #1446 from projectblacklight/remove-and

Allow for multiple extents

app/components/arclight/group_component.html.erb

@@ -7,7 +7,9 @@
         </div>
       <% end %>
       <h3><%= helpers.link_to_document document %></h3>
-      <%= tag.span(document.extent, class: 'al-document-extent badge') if document.extent %>
+      <% document.extent.each do |extent| %>
+        <%= tag.span extent, class: 'al-document-extent badge' unless compact? %>
+      <% end %>        
       <%= render Arclight::IndexMetadataFieldComponent.with_collection(presenter.field_presenters.select { |field| !compact? || field.field_config.compact }) %>
     </div>
   </div>

app/components/arclight/search_result_title_component.html.erb

@@ -1,7 +1,9 @@
 <header class="documentHeader row" data-document-id="<%= @document.id %>">
   <h3 class="index_title document-title-heading col">
     <%= helpers.link_to_document @document, counter: @counter %>
-    <%= tag.span @document.extent, class: 'al-document-extent badge' if @document.extent && !compact? %>
+    <% @document.extent.each do |extent| %>
+      <%= tag.span extent, class: 'al-document-extent badge' unless compact? %>
+    <% end %>    
     <%= tag.span class: 'al-document-container text-muted ml-auto' do %>
       <%= @document.containers.join(', ') %>
     <% end if @document.containers.present? %>

app/models/concerns/arclight/solr_document.rb

@@ -11,7 +11,7 @@ module SolrDocument
       attribute :parent_labels, :array, 'parent_unittitles_ssm'
       attribute :parent_levels, :array, 'parent_levels_ssm'
       attribute :unitid, :string, 'unitid_ssm'
-      attribute :extent, :string, 'extent_ssm'
+      attribute :extent, :array, 'extent_ssm'
       attribute :abstract, :string, 'abstract_html_tesm'
       attribute :scope, :string, 'scopecontent_html_tesm'
       attribute :creator, :string, 'creator_ssm'

lib/arclight/traject/ead2_component_config.rb

@@ -155,8 +155,21 @@
   accumulator.concat settings[:root].output_hash['normalized_title_ssm']
 end
 
-to_field 'extent_ssm', extract_xpath('./did/physdesc/extent')
-to_field 'extent_tesim', extract_xpath('./did/physdesc/extent')
+to_field 'extent_ssm' do |record, accumulator|
+  physdescs = record.xpath('./did/physdesc')
+  extents_per_physdesc = physdescs.map do |physdesc|
+    extents = physdesc.xpath('./extent').map { |e| e.text.strip }
+    # Join extents within the same physdesc with an empty string
+    extents.join(' ')
+  end
+
+  # Add each physdesc separately to the accumulator
+  accumulator.concat(extents_per_physdesc)
+end
+
+to_field 'extent_tesim' do |_record, accumulator, context|
+  accumulator.concat context.output_hash['extent_ssm'] || []
+end
 
 to_field 'creator_ssm', extract_xpath('./did/origination')
 to_field 'creator_ssim', extract_xpath('./did/origination')

lib/arclight/traject/ead2_config.rb

@@ -185,8 +185,22 @@
   end
 end
 
-to_field 'extent_ssm', extract_xpath('/ead/archdesc/did/physdesc/extent')
-to_field 'extent_tesim', extract_xpath('/ead/archdesc/did/physdesc/extent')
+to_field 'extent_ssm' do |record, accumulator|
+  physdescs = record.xpath('/ead/archdesc/did/physdesc')
+  extents_per_physdesc = physdescs.map do |physdesc|
+    extents = physdesc.xpath('./extent').map { |e| e.text.strip }
+    # Join extents within the same physdesc with an empty string
+    extents.join(' ')
+  end
+
+  # Add each physdesc separately to the accumulator
+  accumulator.concat(extents_per_physdesc)
+end
+
+to_field 'extent_tesim' do |_record, accumulator, context|
+  accumulator.concat context.output_hash['extent_ssm'] || []
+end
+
 to_field 'genreform_ssim', extract_xpath('/ead/archdesc/controlaccess/genreform')
 
 to_field 'date_range_isim', extract_xpath('/ead/archdesc/did/unitdate/@normal', to_text: false) do |_record, accumulator|

spec/features/grouped_results_spec.rb

@@ -4,14 +4,14 @@
 
 RSpec.describe 'Grouped search results' do
   it 'displays collection group information' do
-    visit search_catalog_path q: 'alpha omega', group: 'true'
+    visit search_catalog_path q: 'student life', group: 'true'
     within '.al-grouped-title-bar' do
-      expect(page).to have_css 'h3 a', text: /Alpha/
-      expect(page).to have_css '.al-document-abstract-or-scope', text: /founded in 1902/
-      expect(page).to have_css '.badge', text: '15.0 linear feet (36 boxes + oversize folder)'
+      expect(page).to have_css 'h3 a', text: /photograph album/
+      expect(page).to have_css '.al-document-abstract-or-scope', text: /track meets/
+      expect(page).to have_css('.al-document-extent.badge', count: 3)
     end
     within '.grouped-documents' do
-      expect(page).to have_css 'article', count: 3
+      expect(page).to have_css 'article', count: 1
     end
   end
 
@@ -52,7 +52,7 @@
       within '.al-grouped-title-bar' do
         expect(page).to have_css 'h3 a', text: /Alpha/
         expect(page).not_to have_css '.al-document-abstract-or-scope', text: /founded in 1902/
-        expect(page).to have_css '.badge', text: '15.0 linear feet (36 boxes + oversize folder)'
+        expect(page).not_to have_css('.badge')
       end
     end
   end

spec/features/traject/ead2_indexing_spec.rb

@@ -120,6 +120,10 @@ def components(result)
       expect(result['places_ssim']).to equal_array_ignoring_whitespace ['Yosemite National Park (Calif.)']
     end
 
+    it 'physdesc' do
+      expect(result['extent_ssm']).to equal_array_ignoring_whitespace ['1.25 Linear Feet (1 volume)', '1 document case', '16 DVDRs']
+    end
+
     it 'has_online_content' do
       expect(result['has_online_content_ssim']).to eq [true]
     end
@@ -293,9 +297,7 @@ def components(result)
 
     it 'extent at the collection level' do
       %w[extent_ssm extent_tesim].each do |field|
-        expect(result[field]).to equal_array_ignoring_whitespace(
-          ['15.0 linear feet (36 boxes + oversize folder)']
-        )
+        expect(result[field]).to equal_array_ignoring_whitespace(['15.0 linear feet (36 boxes + oversize folder)', '3 CDs'])
       end
     end
 

spec/fixtures/ead/nlm/alphaomegaalpha.xml

@@ -41,10 +41,13 @@
         <corpname source="ingest">Alpha Omega Alpha</corpname>
       </origination>
       <unitid>MS C 271</unitid>
-      <physdesc altrender="whole">
+      <physdesc altrender="part">
         <extent altrender="materialtype spaceoccupied">15.0 linear feet (36 boxes + oversize
           folder)</extent>
       </physdesc>
+      <physdesc altrender="part">
+        <extent altrender="materialtype spaceoccupied">3 CDs</extent>
+      </physdesc>
       <unitdate normal="1894/1992" type="inclusive">1894-1992</unitdate>
       <langmaterial id="aspace_1cf405d4520ad390ab7b5532eab3ea00">Collection materials primarily in
           <language langcode="eng">English.</language></langmaterial>

spec/fixtures/ead/sul-spec/a0011.xml

@@ -77,12 +77,16 @@
                 </corpname>
             </origination>
             <unitid>A0011</unitid>
-            <physdesc altrender="whole">
-                <extent altrender="materialtype spaceoccupied">1.25 Linear
-                    Feet
-                </extent>
+            <physdesc altrender="part">
+                <extent altrender="materialtype spaceoccupied">1.25 Linear Feet</extent>
                 <extent altrender="carrier">(1 volume)</extent>
             </physdesc>
+            <physdesc altrender="part">
+                <extent altrender="materialtype spaceoccupied">1 document case</extent>
+            </physdesc>
+            <physdesc altrender="part">
+                <extent altrender="materialtype spaceoccupied">16 DVDRs</extent>
+            </physdesc>
             <unitdate normal="1900/1906" type="inclusive">circa 1900-1906
             </unitdate>
         </did>

spec/models/arclight/document_downloads_spec.rb

@@ -10,7 +10,6 @@
       id: 'abc123',
       unitid_ssm: ['sample_unitid'],
       userestrict_html_tesm: ['The Terms of the Collection'],
-      extent_ssm: ['42GB'], # This field does not typically hold file size, but for test purposes..
       level_ssm: ['collection'],
       ref_ssm: ['http://example.com/finding-aid.pdf'] # This field does not typically hold URLs, but for test purposes..
     )
@@ -55,10 +54,14 @@
     end
 
     context 'when a size_accessor is provided for the size' do
+      before do
+        allow(document).to receive(:finding_aid_size).and_return('42GB')
+      end
+
       let(:file_params) do
         {
           data: {
-            'size_accessor' => 'extent',
+            'size_accessor' => 'finding_aid_size',
             'href' => 'http://example.com/finding-aid.pdf'
           }
         }