Merge pull request #1521 from sul-dlss/aardvarkDateHierarchy

geo_aardvark_config.rb: add_date_hierarchy_sm field

lib/traject/config/geo_aardvark_config.rb

@@ -174,6 +174,8 @@ def log_skip(context)
 # - used to power the year facet in the UI
 to_field 'gbl_indexYear_im', use_field('dct_temporal_sm'), extract_years, minmax, transform(->(years) { (years.first.to_i..years.last.to_i).to_a if years.any? }), flatten
 
+to_field 'date_hierarchy_sm', use_field('gbl_indexYear_im'), hierarchicalize_year_list
+
 # https://opengeometadata.org/ogm-aardvark/#provider
 to_field 'schema_provider_s', literal('Stanford')
 

lib/traject/macros/extras.rb

@@ -47,6 +47,31 @@ def use_field(field)
           accumulator.concat Array.wrap(context.output_hash[field]) if context.output_hash[field].present?
         end
       end
+
+      # Given a list of years, return a list with the unique centuries and decades covered by those years, and prefixes on the
+      # decade and year strings for easy parsing of century and decade when using the Solr results for hierarchical facet display.
+      # E.g.,
+      # * given: [1701, 1980, 1991, 1995]
+      # * return: ["1700-1799", "1900-1999",
+      #            "1700-1799:1700-1710", "1900-1999:1980-1989", "1900-1999:1990-1999",
+      #            "1700-1799:1700-1710:1701", "1900-1999:1980-1989:1980", "1900-1999:1990-1999:1991", "1900-1999:1990-1999:1995"]
+      # The standalone century and decade ranges make those ranges facetable/searchable, and including the ranges on the year strings
+      # themselves makes it faster for consumers to parse out century/decade info, without duplicating the logic defined in this module.
+      def hierarchicalize_year_list
+        # @param accumulator [Array<Integer>] an array of strings or ints representing calendar years
+        # @return [Array<String>] a list of strings with exploded century and decade info per the above description
+        lambda do |_record, accumulator, _context|
+          centuries = Set.new
+          decades = Set.new
+          hierarchicalized_years = accumulator.map do |year|
+            century, decade = Utils.centimate_and_decimate(year)
+            centuries << century
+            decades << [century, decade].join(':')
+            [century, decade, year].join(':')
+          end
+          accumulator.replace(centuries.to_a + decades.to_a + hierarchicalized_years)
+        end
+      end
     end
   end
 end

lib/utils.rb

@@ -49,6 +49,40 @@ def self.longest_common_call_number_prefix(*strs)
     substr.sub(/(\s|[[:punct:]])+\z/, '')
   end
 
+  # Extract century and decade from year.
+  # "Obsolete. to take a tenth of or from." - https://www.dictionary.com/browse/decimate
+  #
+  # @param maybe_year [Integer] an int that (hopefully) represents a year
+  def self.centimate_and_decimate(maybe_year)
+    parsed_date = Date.new(maybe_year)
+    [century_from_date(parsed_date), decade_from_date(parsed_date)]
+  rescue Date::Error
+    %w[unknown_century unknown_decade] # guess not
+  end
+
+  # Given a Date, return a String for the century that contains it.
+  #
+  # This uses the colloquial grouping of centuries, because it's more intuitive at a glance, and the code is easier:
+  # https://en.wikipedia.org/wiki/Century#Start_and_end_of_centuries
+  #
+  # @param date [Date] a Date object on which we can call strftime
+  # @return [String] a String representing the century in which the date belongs (e.g. 1500-1599)
+  def self.century_from_date(date)
+    date.strftime('%C00-%C99')
+  end
+
+  # Given a Date, return a String for the decade that contains it.
+  #
+  # This uses the more colloquial/popular decade boundary, because it's easier to code and more intuitive for users.
+  # https://en.wikipedia.org/wiki/Decade#0-to-9_decade
+  #
+  # @param date [Date] a Date object on which we can call strftime
+  # @return [String] a String representing the decade in which the date belongs (e.g. 1990-1999)
+  def self.decade_from_date(date)
+    decade_prefix = (date.strftime('%Y').to_i / 10).to_s
+    "#{decade_prefix}0-#{decade_prefix}9"
+  end
+
   def self.version
     @version ||= begin
       file = File.expand_path('../REVISION', __dir__)

spec/integration/geo_aardvark_config_spec.rb

@@ -98,6 +98,14 @@
     expect(result['gbl_indexYear_im']).to eq (1978..2005).to_a
   end
 
+  it 'hierarchicalizes the index years and maps that to a separate field' do
+    expect(result['date_hierarchy_sm'].length).to eq 34 # 28 years, 4 decades, 2 centuries
+    expect(result['date_hierarchy_sm']).to include('1900-1999', '2000-2099',
+                                                   '1900-1999:1970-1979', '1900-1999:1980-1989', '1900-1999:1990-1999',
+                                                   '2000-2099:2000-2009',
+                                                   '1900-1999:1980-1989:1980', '2000-2099:2000-2009:2004')
+  end
+
   it 'maps the provider as stanford' do
     expect(result['schema_provider_s']).to eq 'Stanford'
   end
@@ -347,6 +355,12 @@
       expect(result['gbl_indexYear_im']).to eq [2014]
     end
 
+    it 'hierarchicalizes the index years and maps that to a separate field' do
+      expect(result['date_hierarchy_sm'].sort).to eq(%w[2000-2099
+                                                        2000-2099:2010-2019
+                                                        2000-2099:2010-2019:2014].sort)
+    end
+
     it 'maps the date range' do
       expect(result['gbl_dateRange_drsim']).to eq ['[2014 TO 2014]']
     end
@@ -395,6 +409,13 @@
       expect(result['gbl_indexYear_im']).to eq (1938..1940).to_a
     end
 
+    it 'hierarchicalizes the index years and maps that to a separate field' do
+      expect(result['date_hierarchy_sm'].sort).to eq(%w[1900-1999
+                                                        1900-1999:1930-1939 1900-1999:1940-1949
+                                                        1900-1999:1930-1939:1938 1900-1999:1930-1939:1939
+                                                        1900-1999:1940-1949:1940].sort)
+    end
+
     it 'maps the date range' do
       expect(result['gbl_dateRange_drsim']).to eq ['[1938 TO 1940]']
     end
@@ -463,6 +484,12 @@
       expect(result['gbl_indexYear_im']).to eq [2002]
     end
 
+    it 'hierarchicalizes the index years and maps that to a separate field' do
+      expect(result['date_hierarchy_sm'].sort).to eq(%w[2000-2099
+                                                        2000-2099:2000-2009
+                                                        2000-2099:2000-2009:2002].sort)
+    end
+
     it 'maps the date range' do
       expect(result['gbl_dateRange_drsim']).to eq ['[2002 TO 2002]']
     end