Merge pull request #1495 from projectblacklight/1453-indexentry-tables

Format EAD index/indexentry and EAD table elements as HTML tables. Cl…

app/assets/stylesheets/arclight/modules/layout.scss

@@ -9,7 +9,7 @@
   padding-bottom: $spacer;
   margin-bottom: $spacer * .75;
   width: 100%;
-  
+
   $bent-arrow: "data:image/svg+xml,<svg xmlns='http://www.w3.org/2000/svg' width='20' height='15'><path fill-rule='evenodd' d='M1.5 1.5A.5.5 0 0 0 1 2v4.8a2.5 2.5 0 0 0 2.5 2.5h9.793l-3.347 3.346a.5.5 0 0 0 .708.708l4.2-4.2a.5.5 0 0 0 0-.708l-4-4a.5.5 0 0 0-.708.708L13.293 8.3H3.5A1.5 1.5 0 0 1 2 6.8V2a.5.5 0 0 0-.5-.5z'></path></svg>";
 
   .breadcrumb {
@@ -131,11 +131,11 @@
     }
   }
 
-  svg.bi { 
+  svg.bi {
     vertical-align: text-bottom;
     color: $light-icon-color;
   }
-  
+
   .dropdown-item svg.bi {
     color: $dark-icon-color;
   }
@@ -144,7 +144,7 @@
 .bookmark-toggle {
   display: inline;
   margin-left: 1em;
-  
+
   .toggle-bookmark {
     display: inline;
     margin-bottom: 0;
@@ -177,7 +177,7 @@ dl.deflist dt {
   text-align: left;
 }
 
-.chronlist-head, .list-head {
+.chronlist-head, .list-head, .index-head, .table-head {
   caption-side: top;
   font-size: 1.25rem;
   font-weight: 500;

app/helpers/arclight/ead_format_helpers.rb

@@ -23,20 +23,25 @@ def transform_ead_to_html(value)
             .scrub!(:strip).to_html
     end
 
+    # rubocop:disable Metrics/AbcSize
     def ead_to_html_scrubber
       Loofah::Scrubber.new do |node|
         format_render_attributes(node) if node.attr('render').present?
         convert_to_span(node) if CONVERT_TO_SPAN_TAGS.include? node.name
         convert_to_br(node) if CONVERT_TO_BR_TAG.include? node.name
         format_links(node) if %w[extptr extref extrefloc ptr ref].include? node.name
         format_lists(node) if %w[list chronlist].include? node.name
+        format_indexes(node) if node.name == 'index'
+        format_tables(node) if node.name == 'table'
         node
       end
     end
+    # rubocop:enable Metrics/AbcSize
 
     # Tags that should be converted to <span> tags because of formatting conflicts between XML and HTML
     CONVERT_TO_SPAN_TAGS = ['title'].freeze
-    # Tags that should be converted to <span> tags because of formatting conflicts between XML and HTML
+
+    # Tags that should be converted to HTML <br/> tags
     CONVERT_TO_BR_TAG = ['lb'].freeze
 
     def convert_to_span(node)
@@ -245,5 +250,60 @@ def format_chronlist_events(eventgrps, single_events, multi_events)
       end
       multi_events.each { |event_node| event_node.name = 'div' }
     end
+
+    # Format EAD <index> elements
+    def format_indexes(node)
+      index_head = node.at_css('head')
+      index_head&.name = 'h3'
+      index_head&.add_class('index-head')
+      index_head['id'] = ['index-', index_head.text].join.parameterize if index_head.present?
+      format_indexentries(node)
+      node.name = 'div'
+    end
+
+    # Grab all of the <indexentry> children as a nodeset, move them into
+    # a <table>, wrap each entry in a <tr> & each value in a <td>.
+    def format_indexentries(node)
+      indexentries = node.css('indexentry')
+      return if indexentries.blank?
+
+      indexentries.first.previous = '<table class="table indexentries" />'
+      indexentries.map do |i|
+        i.parent = node.at_css('table.table.indexentries')
+        i.wrap('<tr/>')
+        i.element_children.map { |c| c.wrap('<td/>') }
+        # Assuming two columns in an index, create a blank cell for entries
+        # with a missing value.
+        i.add_child('<td/>') if i.element_children.count == 1
+      end
+    end
+
+    # Format EAD <table> elements, converting <tgroup> to HTML <table>.
+    # Ignoring <colspec>, @colname, @colwidth complex rendering
+    # logic for now.
+    # rubocop:disable Metrics/AbcSize, Metrics/MethodLength
+    def format_tables(node)
+      node.remove_attribute('frame')
+      node.name = 'div' if node.css('tgroup').present?
+      format_table_head(node)
+      tgroups = node.css('tgroup')
+      tgroups&.map do |t|
+        t.name = 'table'
+        t.remove_attribute('cols')
+        t.add_class('table')
+        t.css('row').map { |r| r.name = 'tr' }
+        t.css('thead entry').map { |e| e.name = 'th' }
+        t.css('tbody entry').map { |e| e.name = 'td' }
+      end
+    end
+    # rubocop:enable Metrics/AbcSize, Metrics/MethodLength
+
+    def format_table_head(node)
+      table_head = node.at_css('head')
+      return if table_head.blank?
+
+      table_head.name = 'h3'
+      table_head.add_class('table-head')
+    end
   end
 end

lib/arclight/traject/ead2_component_config.rb

@@ -199,6 +199,9 @@
 to_field 'physfacet_tesim', extract_xpath('./did/physdesc/physfacet')
 to_field 'dimensions_tesim', extract_xpath('./did/physdesc/dimensions')
 
+to_field 'indexes_html_tesm', extract_xpath('./index', to_text: false)
+to_field 'indexes_tesim', extract_xpath('./index')
+
 to_field 'creator_ssm', extract_xpath('./did/origination')
 to_field 'creator_ssim', extract_xpath('./did/origination')
 to_field 'creators_ssim', extract_xpath('./did/origination')

lib/arclight/traject/ead2_config.rb

@@ -226,6 +226,9 @@
   accumulator.replace range.years
 end
 
+to_field 'indexes_html_tesm', extract_xpath('/ead/archdesc/index', to_text: false)
+to_field 'indexes_tesim', extract_xpath('/ead/archdesc/index')
+
 SEARCHABLE_NOTES_FIELDS.map do |selector|
   to_field "#{selector}_html_tesm", extract_xpath("/ead/archdesc/#{selector}/*[local-name()!='head']", to_text: false) do |_record, accumulator|
     accumulator.map!(&:to_html)

lib/generators/arclight/templates/catalog_controller.rb

@@ -327,6 +327,9 @@ class CatalogController < ApplicationController
       last_word_connector: '<br/>'
     }
 
+    config.add_indexed_terms_field 'indexes', field: 'indexes_html_tesm',
+                                              helper_method: :render_html_tags
+
     # ==========================
     # COMPONENT SHOW PAGE FIELDS
     # ==========================
@@ -379,6 +382,9 @@ class CatalogController < ApplicationController
       last_word_connector: '<br/>'
     }
 
+    config.add_component_indexed_terms_field 'indexes', field: 'indexes_html_tesm',
+                                              helper_method: :render_html_tags
+
     # =================
     # ACCESS TAB FIELDS
     # =================

lib/generators/arclight/templates/config/locales/arclight.en.yml

@@ -21,6 +21,7 @@ en:
         physdesc: Physical description
         physfacet: Physical facet
         dimensions: Dimensions
+        indexes: Indexes
         descrules: Rules or conventions
 
         relatedmaterial: Related material

solr/conf/schema.xml

@@ -361,6 +361,7 @@
    <copyField source="custodhist_tesim" dest="text" />
    <copyField source="did_note_tesim" dest="text" />
    <copyField source="fileplan_tesim" dest="text" />
+   <copyField source="indexes_tesim" dest="text" />
    <copyField source="language_ssim" dest="text" />
    <copyField source="materialspec_tesim" dest="text" />
    <copyField source="note_tesim" dest="text" />

spec/features/collection_page_spec.rb

@@ -137,13 +137,15 @@
         expect(page).to have_css('dt', text: 'Subjects')
         expect(page).to have_css('dt', text: 'Names')
         expect(page).to have_css('dt', text: 'Places')
+        expect(page).to have_css('dt', text: 'Indexes')
         expect(page).to have_css('dd', text: 'Societies')
         expect(page).to have_css('dd', text: 'Photographs')
         expect(page).to have_css('dd', text: 'Medicine')
         expect(page).to have_css('dd', text: 'Alpha Omega Alpha')
         expect(page).to have_css('dd', text: 'Root, William Webster, 1867-1932')
         expect(page).to have_css('dd', text: 'Bierring, Walter L. (Walter Lawrence), 1868-1961')
         expect(page).to have_css('dd', text: 'Mindanao Island (Philippines)')
+        expect(page).to have_css('td', text: 'ABC Something')
         expect(page).not_to have_css('dd', text: 'Higgins, L. Raymond')
       end
     end

spec/features/component_page_spec.rb

@@ -92,6 +92,8 @@
       expect(page).to have_css('dd', text: /^various/)
       expect(page).to have_css('dt', text: 'Physical facet')
       expect(page).to have_css('dd', text: /^Boxes and folders/)
+      expect(page).to have_css('dt', text: 'Scope and content')
+      expect(page).to have_css('table thead tr th', text: 'GHI')
     end
 
     it 'multivalued notes are rendered as paragaphs' do

spec/fixtures/ead/nlm/alphaomegaalpha.xml

@@ -385,6 +385,21 @@
       <persname source="mesh">Bierring, Walter L. (Walter Lawrence), 1868-1961</persname>
       <geogname source="lcsh">Mindanao Island (Philippines)</geogname>
     </controlaccess>
+    <index id="aspace_c42daf6e91923b8bee57622436d8f8a4">
+      <head>Alphabetical Listing of Test Titles</head>
+      <indexentry>
+        <title>ABC Something</title>
+        <ref>Box RN2</ref>
+      </indexentry>
+      <indexentry>
+        <title>B402 Something</title>
+        <ref>Box OV2 </ref>
+      </indexentry>
+      <indexentry>
+        <title>C200 Something</title>
+        <ref>Box DO14</ref>
+       </indexentry>
+    </index>
     <dsc>
       <c id="aspace_563a320bb37d24a9e1e6f7bf95b52671" level="series">
         <did>
@@ -501,6 +516,35 @@
             Office in the records storage facility until they were transferred, at his
             family's request in 1988.</p>
           </custodhist>
+          <scopecontent id="aspace_63fb9d2df5fb62ff01de34b0dd002155">
+            <p>Here is a table in a paragraph.</p>
+            <p>
+              <table frame="none">
+                <head>Test Table</head>
+                <tgroup cols="3">
+                  <thead>
+                    <row>
+                      <entry colname="1">ABC</entry>
+                      <entry colname="2">DEF</entry>
+                      <entry colname="3">GHI</entry>
+                    </row>
+                  </thead>
+                  <tbody>
+                    <row>
+                      <entry colname="1">123</entry>
+                      <entry colname="2">456</entry>
+                      <entry colname="3">789</entry>
+                    </row>
+                    <row>
+                      <entry colname="1">012</entry>
+                      <entry colname="2">345</entry>
+                      <entry colname="3">678</entry>
+                    </row>
+                  </tbody>
+                </tgroup>
+              </table>
+            </p>
+          </scopecontent>
         </c>
         <c id="aspace_a8b5c3a57013462581d0e807241fcf93" level="otherlevel">
           <did>

spec/helpers/arclight/ead_format_helpers_spec.rb

@@ -425,5 +425,97 @@
         end
       end
     end
+
+    describe 'indexes' do
+      describe 'basic index with name/ref pairs' do
+        it 'index -> table' do
+          content = helper.render_html_tags(value: [%(
+            <index>
+              <head>Alphabetical Listing By Client</head>
+              <p>Here's a brief note describing the index</p>
+              <indexentry>
+                <name>Absorbine Jr.</name>
+                <ref>1947 Feb. 17</ref>
+              </indexentry>
+              <indexentry>
+                <name>Action Gemeinsinn</name>
+                <ref>1962 Feb. 7</ref>
+              </indexentry>
+              <indexentry>
+                <name>Alexander Smith Inc.</name>
+                <ref>1952 Mar. 24</ref>
+              </indexentry>
+            </index>
+          )])
+          expect(content).to eq_ignoring_whitespace %(
+            <div>
+              <h3 class="index-head" id="index-alphabetical-listing-by-client">Alphabetical Listing By Client</h3>
+              <p>Here's a brief note describing the index</p>
+              <table class="table indexentries">
+                <tr>
+                 <td>Absorbine Jr.</td>
+                 <td>1947 Feb. 17</td>
+                </tr>
+                <tr>
+                  <td>Action Gemeinsinn</td>
+                  <td>1962 Feb. 7</td>
+                </tr>
+                <tr>
+                  <td>Alexander Smith Inc.</td>
+                  <td>1952 Mar. 24</td>
+                </tr>
+              </table>
+            </div>
+          )
+        end
+      end
+    end
+
+    describe 'tables' do
+      describe 'table with headers' do
+        it 'EAD table -> HTML table' do
+          content = helper.render_html_tags(value: [%(
+            <table frame="none">
+              <tgroup cols="3">
+                <thead>
+                  <row>
+                    <entry colname="1">Major Family Members</entry>
+                    <entry colname="2">Spouses</entry>
+                    <entry colname="3">Children</entry>
+                  </row>
+                </thead>
+                <tbody>
+                  <row>
+                    <entry colname="1">John Albemarle (1760-1806)</entry>
+                    <entry colname="2">Mary Frances Delaney (1769-1835)</entry>
+                    <entry colname="3">John Delaney Albemarle (1787-1848)</entry>
+                  </row>
+                </tbody>
+              </tgroup>
+            </table>
+          )])
+          expect(content).to eq_ignoring_whitespace %(
+            <div>
+              <table class="table">
+                <thead>
+                  <tr>
+                    <th>Major Family Members</th>
+                    <th>Spouses</th>
+                    <th>Children</th>
+                  </tr>
+                </thead>
+                <tbody>
+                  <tr>
+                    <td>John Albemarle (1760-1806)</td>
+                    <td>Mary Frances Delaney (1769-1835)</td>
+                    <td>John Delaney Albemarle (1787-1848)</td>
+                  </tr>
+                </tbody>
+              </table>
+            </div>
+          )
+        end
+      end
+    end
   end
 end