Add catch-some fields for key bib data

lib/traject/config/folio_config.rb

@@ -368,6 +368,25 @@ def holdings(record, context)
          extract_marc("440anpv:490av:800#{A_X}:810#{A_X}:811#{A_X}:830#{A_X}", alternate_script: :only)
 to_field 'series_exact_search', extract_marc('830a', alternate_script: false)
 
+to_field 'author_title_245ac_search', extract_marc('245ac', alternate_script: false)
+to_field 'vern_author_title_245ac_search', extract_marc('245ac', alternate_script: :only)
+
+to_field 'author_title_1xx_search' do |record, accumulator|
+  onexx = trim_punctuation_when_preceded_by_two_word_characters_or_some_other_stuff(Traject::MarcExtractor.cached('100abcq:110abn:111aeq', alternate_script: false).extract(record).first)
+  twoxx = trim_punctuation_when_preceded_by_two_word_characters_or_some_other_stuff(Traject::MarcExtractor.cached('240' + ALPHABET, alternate_script: false).extract(record).first)
+  twoxx ||= Traject::MarcExtractor.cached('245aa', alternate_script: false).extract(record).first
+
+  accumulator << [onexx, twoxx].compact.reject(&:empty?).map(&:strip).join(' ') if onexx
+end
+
+to_field 'vern_author_title_1xx_search' do |record, accumulator|
+  onexx = trim_punctuation_when_preceded_by_two_word_characters_or_some_other_stuff(Traject::MarcExtractor.cached('100abcq:110abn:111aeq', alternate_script: :only).extract(record).first)
+  twoxx = trim_punctuation_when_preceded_by_two_word_characters_or_some_other_stuff(Traject::MarcExtractor.cached('240' + ALPHABET, alternate_script: :only).extract(record).first)
+  twoxx ||= Traject::MarcExtractor.cached('245aa', alternate_script: :only).extract(record).first
+
+  accumulator << [onexx, twoxx].compact.reject(&:empty?).map(&:strip).join(' ') if onexx
+end
+
 # # Author Title Search Fields
 to_field 'author_title_search' do |record, accumulator|
   onexx = trim_punctuation_when_preceded_by_two_word_characters_or_some_other_stuff(Traject::MarcExtractor.cached(
@@ -380,7 +399,7 @@ def holdings(record, context)
   twoxx ||= Traject::MarcExtractor.cached('245aa', alternate_script: false).extract(record).first if record['245']
   twoxx ||= 'null'
 
-  accumulator << [onexx, twoxx].compact.reject(&:empty?).map(&:strip).join(' ') if onexx
+  accumulator << [onexx, twoxx].compact.reject(&:empty?).map(&:strip).join(' ') if onexx && twoxx
 end
 
 to_field 'author_title_search' do |record, accumulator|
@@ -389,7 +408,6 @@ def holdings(record, context)
   ).extract(record).first
 
   twoxx = Traject::MarcExtractor.cached('240' + ALPHABET, alternate_script: :only).extract(record).first
-  twoxx ||= Traject::MarcExtractor.cached('245aa', alternate_script: :only).extract(record).first
   accumulator << [onexx, twoxx].compact.reject(&:empty?).map(&:strip).join(' ') if onexx && twoxx
 end
 
@@ -413,6 +431,21 @@ def holdings(record, context)
   end
 end
 
+to_field 'author_title_search' do |record, accumulator|
+  onexx = trim_punctuation_when_preceded_by_two_word_characters_or_some_other_stuff(Traject::MarcExtractor.cached('100abcdfghijklmnopqrstuvwxyz:110abcdfghijklmnopqrstuvwxyz:111abcdefghjklmnopqrstuvwxyz',
+                                                                                                                  alternate_script: false).extract(record).first)
+
+  twoxx = Traject::MarcExtractor.cached('245aa', alternate_script: false).extract(record).first if record['245']
+  accumulator << [onexx, twoxx].compact.reject(&:empty?).map(&:strip).join(' ') if onexx && twoxx
+end
+
+to_field 'author_title_search' do |record, accumulator|
+  onexx = Traject::MarcExtractor.cached('100abcdfghijklmnopqrstuvwxyz:110abcdfghijklmnopqrstuvwxyz:111abcdefghjklmnopqrstuvwxyz', alternate_script: :only).extract(record).first
+
+  twoxx = Traject::MarcExtractor.cached('245aa', alternate_script: :only).extract(record).first
+  accumulator << [onexx, twoxx].compact.reject(&:empty?).map(&:strip).join(' ') if onexx && twoxx
+end
+
 # # Author Search Fields
 # # IFF relevancy of author search needs improvement, unstemmed flavors for author search
 # #   (keep using stemmed version for everything search to match stemmed query)
@@ -2575,6 +2608,55 @@ def accumulate_summary_struct_fields(matching_fields, tag, label, marc, accumula
   end.flatten.uniq)
 end
 
+to_field 'bib_search' do |record, accumulator, context|
+  # authors, titles, series, publisher
+  keep_fields = %w[
+    100 110 111 130 210 222 242 243 245 246 247 260 264 440 490 700 710 711 800 810 811
+  ]
+
+  result = []
+  record.each do |field|
+    next unless keep_fields.include?(field.tag)
+
+    subfield_values = field.subfields
+                           .reject { |sf| Constants::EXCLUDE_FIELDS.include?(sf.code) }
+                           .collect(&:value)
+
+    next unless subfield_values.length > 0
+
+    result << subfield_values.join(' ')
+  end
+
+  result += Array(context.output_hash['format_main_ssim'])
+
+  accumulator << result.join(' ') if result.any?
+end
+
+to_field 'vern_bib_search' do |record, accumulator|
+  # authors, titles, series, publisher
+  keep_fields = %w[
+    100 110 111 130 210 222 242 243 245 246 247 260 264 440 490 700 710 711 800 810 811
+  ]
+
+  result = []
+  record.each do |field|
+    next unless field.tag == '880'
+    next if field['6'].nil? ||
+            !field['6'].include?('-') ||
+            !keep_fields.include?(field['6'].split('-')[0])
+
+    subfield_values = field.subfields
+                           .reject { |sf| Constants::EXCLUDE_FIELDS.include?(sf.code) }
+                           .collect(&:value)
+
+    next unless subfield_values.length > 0
+
+    result << subfield_values.join(' ')
+  end
+
+  accumulator << result.join(' ') if result.any?
+end
+
 ## FOLIO specific fields
 
 ## QUESTIONS / ISSUES

spec/integration/compare_against_solrmarc_docs_spec.rb

@@ -15,7 +15,7 @@
        title_variant_display summary_display pub_display]
   end
   let(:ignored_fields) do
-    %w[all_search created last_updated format _version_ author_sort callnum_facet_hsim marcbib_xml marcxml mhld_display fund_facet building_facet collection] + copy_fields
+    %w[all_search created last_updated format _version_ author_sort author_title_search callnum_facet_hsim marcbib_xml marcxml mhld_display fund_facet building_facet collection] + copy_fields
   end
   let(:pending_fields) { %w[reverse_shelfkey shelfkey preferred_barcode item_display date_cataloged access_facet] }
   subject(:result) { indexer.map_record(folio_record).transform_values { |v| v.sort } }

spec/lib/traject/config/all_search_spec.rb

@@ -37,4 +37,12 @@
       # all_search should include 033a
     end
   end
+  describe 'bib_search' do
+    let(:field) { 'bib_search' }
+    it do
+      expect(select_by_id('allfields1')[field]).to include(/Dharma Kumar/)
+      expect(select_by_id('allfields1')[field]).to include(/allfields should skip 00x/)
+      expect(select_by_id('allfields1')[field]).to include(%r{Journal/Periodical})
+    end
+  end
 end

spec/lib/traject/config/author_title_spec.rb

@@ -16,9 +16,18 @@
   # rubocop:disable Layout/LineLength
   describe 'maps search field values from 100, 110, 111 with data from the 240 or 245' do
     it 'maps the right data' do
-      expect(select_by_id('100240')[field]).to eq ['100a 100b 100c 100d 100f 100g 100j 100k 100l 100n 100p 100q 100t 100u 240a 240d 240f 240g 240h 240k 240l 240m 240n 240o 240p 240r 240s']
-      expect(select_by_id('110240')[field]).to eq ['110a 110b 110c 110d 110f 110g 110k 110l 110n 110p 110t 110u 240a 240d 240f 240g 240h 240k 240l 240m 240n 240o 240p 240r 240s']
-      expect(select_by_id('111240')[field]).to eq ['111a 111c 111d 111e 111f 111g 111j 111k 111l 111n 111p 111q 111t 111u 240a 240d 240f 240g 240h 240k 240l 240m 240n 240o 240p 240r 240s']
+      expect(select_by_id('100240')[field]).to eq [
+        '100a 100b 100c 100d 100f 100g 100j 100k 100l 100n 100p 100q 100t 100u 240a 240d 240f 240g 240h 240k 240l 240m 240n 240o 240p 240r 240s',
+        '100a 100b 100c 100d 100f 100g 100j 100k 100l 100n 100p 100q 100t 100u 100 all subfields + 240 all subfields'
+      ]
+      expect(select_by_id('110240')[field]).to eq [
+        '110a 110b 110c 110d 110f 110g 110k 110l 110n 110p 110t 110u 240a 240d 240f 240g 240h 240k 240l 240m 240n 240o 240p 240r 240s',
+        '110a 110b 110c 110d 110f 110g 110k 110l 110n 110p 110t 110u 110 all subfields + 240 all subfields'
+      ]
+      expect(select_by_id('111240')[field]).to eq [
+        '111a 111c 111d 111e 111f 111g 111j 111k 111l 111n 111p 111q 111t 111u 240a 240d 240f 240g 240h 240k 240l 240m 240n 240o 240p 240r 240s',
+        '111a 111c 111d 111e 111f 111g 111j 111k 111l 111n 111p 111q 111t 111u 111 all subfields + 240 all subfields'
+      ]
 
       expect(select_by_id('100no240')[field]).to eq ['100a 100b 100c 100d 100f 100g 100j 100k 100l 100n 100p 100q 100t 100u 245a']
       expect(select_by_id('110no240')[field]).to eq ['110a 110b 110c 110d 110f 110g 110k 110l 110n 110p 110t 110u 245a']
@@ -62,7 +71,7 @@
 
     it 'does something with a minimal 100 field to link to 880' do
       expect(select_by_id('vern100no240')[field]).to eq ['100a 245a']
-      expect(select_by_id('vern100plain240')[field]).to eq ['100a 240a 240d 240f 240g 240h 240k 240l 240m 240n 240o 240p 240r 240s']
+      expect(select_by_id('vern100plain240')[field]).to eq ['100a 240a 240d 240f 240g 240h 240k 240l 240m 240n 240o 240p 240r 240s', '100a 245a']
 
       expect(select_by_id('vern110vern240')[field]).to include 'vern110a vern110b vern110c vern110d vern110f vern110g vern110k vern110l vern110n vern110p vern110t vern110u vern240a vern240d vern240f vern240g vern240h vern240k vern240l vern240m vern240n vern240o vern240p vern240r vern240s'
       expect(select_by_id('vern110vern245')[field]).to include 'vern110a vern110b vern110c vern110d vern110f vern110g vern110k vern110l vern110n vern110p vern110t vern110u vern245a'