Base call for number for Dewey classifications. (#353)

Using the Lcsort gem, creates shelf keys for dewey numbers to sort
in forward and reverse directions.

.rubocop.yml

@@ -12,6 +12,8 @@ inherit_from:
 AllCops:
   DisplayCopNames: true
   NewCops: enable
+  Exclude:
+    - 'config/ignorethis_*.rb'
 
 Lint/ConstantDefinitionInBlock:
   Exclude:
@@ -35,7 +37,18 @@ Lint/MissingSuper:
 Lint/MixedRegexpCaptureTypes:
   Exclude:
     - 'lib/psulib_traject/processors/call_number/lc.rb'
+    - 'lib/psulib_traject/processors/call_number/dewey.rb'
 
 RSpec/ImplicitSubject:
   Exclude:
     - 'spec/lib/psulib_traject/macros/subjects_spec.rb'
+
+RSpec/NestedGroups:
+  Exclude:
+    - 'spec/lib/psulib_traject/processors/call_number/lc_spec.rb'
+    - 'spec/lib/psulib_traject/processors/call_number/dewey_spec.rb'
+
+RSpec/ExpectActual:
+  Exclude:
+    - 'spec/lib/psulib_traject/processors/call_number/lc_spec.rb'
+    - 'spec/lib/psulib_traject/processors/call_number/dewey_spec.rb'

.rubocop_todo.yml

@@ -1,42 +1,31 @@
 # This configuration was generated by
-# `rubocop --auto-gen-config --auto-gen-only-exclude`
-# on 2021-09-07 19:44:42 UTC using RuboCop version 0.93.1.
+# `rubocop --auto-gen-config`
+# on 2021-09-10 20:26:43 UTC using RuboCop version 0.93.1.
 # The point is for the user to remove these configuration records
 # one by one as the offenses are removed from the code base.
 # Note that changes in the inspected code, or installation of new
 # versions of RuboCop, may require this file to be generated again.
 
 # Offense count: 4
-# Configuration parameters: CountComments, Max, CountAsOne, ExcludedMethods.
+# Configuration parameters: CountComments, CountAsOne, ExcludedMethods.
 # ExcludedMethods: refine
 Metrics/BlockLength:
-  Exclude:
-    - 'spec/**/*'
-    - 'lib/tasks/docker.rake'
-    - 'lib/tasks/incrementals.rake'
+  Max: 75
 
-# Offense count: 1
-# Configuration parameters: CountComments, Max, CountAsOne.
+# Offense count: 2
+# Configuration parameters: CountComments, CountAsOne.
 Metrics/ClassLength:
-  Exclude:
-    - 'lib/psulib_traject/processors/media_type.rb'
+  Max: 138
 
 # Offense count: 4
-# Configuration parameters: IgnoredMethods, Max.
+# Configuration parameters: IgnoredMethods.
 Metrics/CyclomaticComplexity:
-  Exclude:
-    - 'lib/psulib_traject/holdings.rb'
-    - 'lib/psulib_traject/macros/subjects.rb'
-    - 'lib/psulib_traject/processors/media_type.rb'
-    - 'lib/psulib_traject/processors/preferred_format.rb'
+  Max: 10
 
 # Offense count: 3
-# Configuration parameters: IgnoredMethods, Max.
+# Configuration parameters: IgnoredMethods.
 Metrics/PerceivedComplexity:
-  Exclude:
-    - 'lib/psulib_traject/holdings.rb'
-    - 'lib/psulib_traject/macros/subjects.rb'
-    - 'lib/psulib_traject/marc_combining_reader.rb'
+  Max: 10
 
 # Offense count: 2
 # Configuration parameters: MinSize.
@@ -60,49 +49,19 @@ RSpec/DescribeClass:
     - 'spec/integration/access_facet_spec.rb'
     - 'spec/integration/bound_with_spec.rb'
 
-# Offense count: 56
-# Cop supports --auto-correct.
-RSpec/ExpectActual:
-  Exclude:
-    - 'spec/routing/**/*'
-    - 'spec/lib/psulib_traject/processors/call_number/lc_spec.rb'
-
-# Offense count: 20
-# Configuration parameters: AllowSubject, Max.
+# Offense count: 23
+# Configuration parameters: AllowSubject.
 RSpec/MultipleMemoizedHelpers:
-  Exclude:
-    - 'spec/integration/macros_spec.rb'
-    - 'spec/integration/title_spec.rb'
-    - 'spec/lib/psulib_traject/holdings_spec.rb'
-
-# Offense count: 2
-# Configuration parameters: Max.
-RSpec/NestedGroups:
-  Exclude:
-    - 'spec/lib/psulib_traject/processors/call_number/lc_spec.rb'
+  Max: 9
 
 # Offense count: 4
 Style/MixinUsage:
   Exclude:
     - 'config/traject.rb'
 
-# Offense count: 59
+# Offense count: 72
 # Cop supports --auto-correct.
-# Configuration parameters: AutoCorrect, Max, AllowHeredoc, AllowURI, URISchemes, IgnoreCopDirectives, IgnoredPatterns.
+# Configuration parameters: AutoCorrect, AllowHeredoc, AllowURI, URISchemes, IgnoreCopDirectives, IgnoredPatterns.
 # URISchemes: http, https
 Layout/LineLength:
-  Exclude:
-    - 'config/schedule.rb'
-    - 'config/traject.rb'
-    - 'lib/psulib_traject/hathi_overlap_reducer.rb'
-    - 'lib/psulib_traject/macros.rb'
-    - 'lib/psulib_traject/processors/call_number/base.rb'
-    - 'lib/psulib_traject/processors/media_type.rb'
-    - 'lib/psulib_traject/processors/pub_date.rb'
-    - 'spec/integration/bound_with_spec.rb'
-    - 'spec/integration/config_spec.rb'
-    - 'spec/integration/macros_spec.rb'
-    - 'spec/integration/title_spec.rb'
-    - 'spec/lib/psulib_traject/holdings_spec.rb'
-    - 'spec/lib/psulib_traject/marc_combining_reader_spec.rb'
-    - 'spec/lib/psulib_traject/processors/pub_date_spec.rb'
+  Max: 231

Gemfile.lock

@@ -246,6 +246,7 @@ GEM
     yell (2.2.2)
 
 PLATFORMS
+  universal-java-10
   universal-java-11
   universal-java-14
   x86_64-darwin-20

config/traject.rb

@@ -293,14 +293,19 @@
   accumulator.replace [lc_rest]
 end
 
-# Determines a base call number from the record's holdings and creates forward and reverse shelfkeys
+# Call Number Browse
+#
+## Determines a base call number from the record's holdings and creates forward and reverse shelfkeys for LC, LCPER and DEWEY
 each_record do |record, context|
-  call_numbers = PsulibTraject::Holdings.call(record: record, context: context, classification: ['LC', 'LCPER'])
+  call_numbers = PsulibTraject::Holdings.call(record: record, context: context)
   next if call_numbers.empty?
 
-  context.add_output('call_number_lc_ssm', *call_numbers.map(&:value))
-  context.add_output('forward_lc_shelfkey', *call_numbers.map(&:forward_shelfkey))
-  context.add_output('reverse_lc_shelfkey', *call_numbers.map(&:reverse_shelfkey))
+  call_numbers.each do |call_number|
+    context.add_output(call_number.solr_field, call_number.value)
+    context.add_output(call_number.forward_shelfkey_field, call_number.forward_shelfkey)
+    context.add_output(call_number.reverse_shelfkey_field, call_number.reverse_shelfkey)
+  end
+
   context.add_output('keymap_struct', *call_numbers.map(&:keymap).to_json)
 end
 

lib/psulib_traject/call_number.rb

@@ -10,6 +10,8 @@ class CallNumber
       BNDSER-HY
     ).freeze
 
+    DEWEY_KLASS_PREFIX = 'AAA'
+
     attr_reader :value, :classification, :location, :item_type, :leader
 
     def initialize(value: '', classification: '', location: '', item_type: '', leader: '')
@@ -32,11 +34,18 @@ def reverse_shelfkey
       shelf_key.reverse
     end
 
+    def not_browsable?
+      return true unless lc? || dewey?
+
+      forward_shelfkey.nil? && reverse_shelfkey.nil?
+    end
+
     def keymap
       {
-        'call_number' => value,
-        'forward_key' => forward_shelfkey,
-        'reverse_key' => reverse_shelfkey
+        call_number: value,
+        classification: classification,
+        forward_key: forward_shelfkey,
+        reverse_key: reverse_shelfkey
       }
     end
 
@@ -65,21 +74,54 @@ def serial?
         item_type == 'MICROFORM' && %w(ab as).include?(leader[6..7])
     end
 
+    def solr_field
+      "call_number_#{classification_to_field}_ssm"
+    end
+
+    def forward_shelfkey_field
+      "forward_#{classification_to_field}_shelfkey"
+    end
+
+    def reverse_shelfkey_field
+      "reverse_#{classification_to_field}_shelfkey"
+    end
+
     private
 
       def base_value
         case classification
         when 'LC', 'LCPER'
           PsulibTraject::Processors::CallNumber::LC.new(value, serial: serial?).reduce
         when 'DEWEY'
-          PsulibTraject::Processors::CallNumber::Dewey.new(value).reduce
+          PsulibTraject::Processors::CallNumber::Dewey.new(value, serial: serial?).reduce
         else
           PsulibTraject::Processors::CallNumber::Other.new(value).reduce
         end
       end
 
+      def lc?
+        %w[LC LCPER].include? classification
+      end
+
+      def dewey?
+        classification == 'DEWEY'
+      end
+
+      def classification_to_field
+        return 'lc' if lc?
+
+        classification.downcase
+      end
+
+      # @note Adding a prefix to dewey call number so to be able to use lcsort to create shelf keys
+      def prefix
+        return '' unless dewey?
+
+        DEWEY_KLASS_PREFIX
+      end
+
       def shelf_key
-        @shelf_key ||= ShelfKey.new(value)
+        @shelf_key ||= ShelfKey.new(value, prefix: prefix)
       end
   end
 end

lib/psulib_traject/holdings.rb

@@ -4,31 +4,29 @@ module PsulibTraject
   class Holdings
     # @param record [Marc::Record]
     # @param context [Traject::Indexer::Context]
-    # @param classification [Array<String>] Returns call numbers only for given classifications, example: 'LC', 'LCPER',
-    #   'DEWEY'. Defaults to [], which return any classification.
     # @return [Array<CallNumber>]
-    def self.call(record:, context:, classification: [])
+    def self.call(record:, context:)
       new(
         record: record,
-        context: context,
-        classification: Array(classification)
+        context: context
       ).resolve_base
     end
 
-    attr_reader :record, :context, :holdings, :classification
+    attr_reader :record, :context, :holdings
 
-    def initialize(record:, context:, classification:)
+    def initialize(record:, context:)
       @record = record
       @context = context
       @holdings = extract_holdings
-      @classification = classification
       freeze
     end
 
     def resolve_base
       return [] if online? || holdings.empty?
 
-      holdings.reject! { |call_number| call_number.exclude? || classification_not_requested?(call_number) }
+      holdings.reject! do |call_number|
+        call_number.exclude? || call_number.not_browsable?
+      end
 
       if holdings.one?
         holdings
@@ -45,12 +43,6 @@ def online?
         context.output_hash['access_facet']&.include?('Online')
       end
 
-      def classification_not_requested?(call_number)
-        return false if classification.empty?
-
-        !classification.include?(call_number.classification)
-      end
-
       # assuming each 949 has only one subfield a, w and l
       def extract_holdings
         Traject::MarcExtractor.cached('949').collect_matching_lines(record) do |field, _spec, _extractor|

lib/psulib_traject/processors/call_number/base.rb

@@ -23,7 +23,7 @@ class Base
       shanah
       tbd
     ).join('|')
-      .concat("[^a-z]t|v|vols?|vyp|k\\.|h\\.|ḥ\\.|#{ORDINALS}")
+      .concat("[^a-z]t|v|vols?|vyp|k\\.|h\\.|ḥ\\.|t\\.|#{ORDINALS}")
 
     ADDL_VOL_PARTS = %w(
       bklet
@@ -60,15 +60,25 @@ class Base
 
     ADDL_VOL_PATTERN = /[:\/]?(#{ADDL_VOL_PARTS.join('|')}).*/i.freeze
     VOL_PARTS_ALL = "((index|ind)\s)?(#{VOL_PARTS}|#{MONTHS})"
-    VOL_PATTERN         = /([.:\/(])?(n\.s\.?,? ?)?[:\/]?#{VOL_PARTS_ALL}[. -\/]?\d+([\/-]\d+)?( \d{4}([\/-]\d{4})?)?( ?suppl\.?)?/i.freeze
+    VOL_PATTERN         = /([.:\/(])?(n\.s\.?,? ?)?[:\/]?#{VOL_PARTS_ALL}[. \-\/]?\d+([\/-]\d+)?( \d{4}([\/-]\d{4})?)?( ?suppl\.?)?/i.freeze
     VOL_PATTERN_LOOSER  = /([.:\/(])?(n\.s\.?,? ?)?[:\/]?#{VOL_PARTS_ALL}[. -]?\d+.*/i.freeze
     VOL_PATTERN_LETTERS = /([.:\/(])?(n\.s\.?,? ?)?[:\/]?#{VOL_PARTS_ALL}[\/. -]?[A-Z]?([\/-][A-Z]+)?.*/i.freeze
     FOUR_DIGIT_YEAR_REGEX = /\W *(20|19|18|17|16|15|14)\d{2}\D?$?/.freeze
     LOOSE_MONTHS_REGEX = /([.:\/(])? *#{MONTHS}/i.freeze
 
-    def reduce
-      raise NotImplementedError
-    end
+    attr_reader :call_number,
+                :cutter1,
+                :cutter2,
+                :cutter3,
+                :doon1,
+                :doon2,
+                :doon3,
+                :klass,
+                :klass_decimal,
+                :klass_number,
+                :removeables,
+                :rest,
+                :serial
 
     class << self
       def remove_years(value)
@@ -80,5 +90,44 @@ def remove_years(value)
         shortest_value
       end
     end
+
+    def reduce
+      value = remove_by_regex
+      value = value[0...(value.index(LOOSE_MONTHS_REGEX) || value.length)] # remove loose months
+
+      if serial
+        self.class.remove_years(value)
+      else
+        value.strip
+      end
+    end
+
+    private
+
+      # @note These are the original regex patterns from Stanford. However, VOL_PATTERN_LOOSER does not currently apply
+      # to any of our test data, so it has been commented-out of the procedure.
+      def remove_by_regex
+        case removeables
+        when VOL_PATTERN
+          call_number.slice(0...call_number.index(removeables[VOL_PATTERN])).strip
+        # when VOL_PATTERN_LOOSER
+        #   call_number.slice(0...call_number.index(removeables[VOL_PATTERN_LOOSER])).strip
+        when /Blu-ray|DVD/
+          bluray_or_dvd
+        when VOL_PATTERN_LETTERS
+          call_number.slice(0...call_number.index(removeables[VOL_PATTERN_LETTERS])).strip
+        when ADDL_VOL_PATTERN
+          call_number.slice(0...call_number.index(removeables[ADDL_VOL_PATTERN])).strip
+        else
+          call_number
+        end
+      end
+
+      def bluray_or_dvd
+        element = removeables[ADDL_VOL_PATTERN]
+        return call_number unless element
+
+        call_number.slice(0...call_number.index(removeables[ADDL_VOL_PATTERN])).strip
+      end
   end
 end