pennmarc 1.0.15 → 1.0.17

checksums.yaml

@@ -1,7 +1,7 @@
 ---
 SHA256:
-  metadata.gz: 06f7c00af4fabc0cb127ce3bfc16e02e2b13611db0aa8184fca48392a5776295
-  data.tar.gz: c34a742ed5b5bc1f1eb1d76a564738e0e48df3607e59ec69851a7428876225cb
+  metadata.gz: 90b9c248dbc3c08377212a70748dcf13728c9feaa037a4a64cc398c02a36c3df
+  data.tar.gz: f3c14ffea869113e6f94fc5d94ff5a083f2749cb17ed33854243131b134185a1
 SHA512:
-  metadata.gz: f64552e53b3547222f785bf3f24f593448a3e4587baa309d194ca48fee961d9846b40c590b4e2a8709592557c95a0c52b922c5c4fe9be89058cc85aefe2c2e76
-  data.tar.gz: 999d8a04a067bf9a1a88d4d0adba0947fda547446b89d5f989541ea18c087d0ab804dde297f67704cbfcd6ca5be4047db4296ffb87f4e68b9c81e991b954c316
+  metadata.gz: 25f98a210e26abc52edefb5614bf5ebbdb1acc04baaaaa3c0a770f284f8161f301a2fe768a336ee9291ba640f9251e71d68d39cd5d222707254b8ca5763261b6
+  data.tar.gz: c0ee733d64114e1b5c317cd49ebedc330c2a179282be2cb43f3f9457d40253dd197da5fe18e6c09a04f945be1d43fa837a9920648ae8af439a3ad0fdcf23ee38

data/lib/pennmarc/heading_control.rb

@@ -6,6 +6,6 @@ module PennMARC
     # These codes are expected to be found in sf2 when the indicator2 value is 7, indicating "source specified". There
     # are some sources whose headings we don't want to display.
     ALLOWED_SOURCE_CODES = %w[aat cct fast ftamc gmgpc gsafd homoit jlabsh lcgft lcsh lcstt lctgm
-                              local/osu mesh ndlsh nlksh rbbin rbgenr rbmscv rbpap rbpri rbprov rbpub rbtyp].freeze
+                              local/osu mesh ndlsh nli nlksh rbbin rbgenr rbmscv rbpap rbpri rbprov rbpub rbtyp].freeze
   end
 end

data/lib/pennmarc/helpers/access.rb

@@ -16,15 +16,16 @@ module PennMARC
       # @param [MARC::Record] record
       # @return [Array]
       def facet(record)
-        acc = record.filter_map do |field|
+        values = record.filter_map do |field|
           next AT_THE_LIBRARY if physical_holding_tag?(field)
           next ONLINE if electronic_holding_tag?(field)
         end
 
-        return acc if acc.size == 2 # return early if all values are already present
+        return values if values.size == 2 # return early if all values are already present
 
-        acc << ONLINE if acc.exclude?(ONLINE) && finding_aid_linkage?(record) # only check if ONLINE isn't already there
-        acc
+        # only check if ONLINE isn't already there
+        values << ONLINE if values.exclude?(ONLINE) && finding_aid_linkage?(record)
+        values.uniq
       end
 
       private

data/lib/pennmarc/helpers/format.rb

@@ -27,7 +27,7 @@ module PennMARC
       WEBSITE_DATABASE = 'Website/Database'
 
       # Get any Format values from {https://www.oclc.org/bibformats/en/3xx/300.html 300},
-      # 254, 255, 310, 342, 352 or {https://www.oclc.org/bibformats/en/3xx/340.html 340} field. based on the source
+      # 254, 255, 310, 342, 352, 362 or {https://www.oclc.org/bibformats/en/3xx/340.html 340} field. based on the source
       # field, different subfields are used.
       # @note ported from get_format_display
       # @param [MARC::Record] record
@@ -39,6 +39,9 @@ module PennMARC
         end
         results += record.fields('340').map { |f| join_subfields(f, &subfield_not_in?(%w[0 2 6 8])) }
         results += record.fields('880').map do |f|
+          # skip any 880s associated with non format fields
+          next unless subfield_value_in?(f, '6', %w[254 255 300 310 340 342 352 362])
+
           subfield_to_ignore = if subfield_value?(f, 6, /^300/)
                                  %w[3 6 8]
                                elsif subfield_value?(f, 6, /^340/)

data/lib/pennmarc/helpers/location.rb

@@ -4,6 +4,7 @@ module PennMARC
   # Methods that return Library and Location values from Alma enhanced MARC fields
   class Location < Helper
     ONLINE_LIBRARY = 'Online library'
+    WEB_LOCATION_CODE = 'web'
 
     class << self
       # Retrieves library location from enriched marc 'itm' or 'hld' fields, giving priority to the item location over
@@ -41,43 +42,45 @@ module PennMARC
       # @param [Hash] location_map hash with location_code as key and location hash as value
       # @return [Array<String>]
       def location(record:, display_value:, location_map:)
-        # get enriched marc location tag and subfield code
-        location_tag_and_subfield_code(record) => {tag:, subfield_code:}
+        # get enriched marc location tag and relevant subfields
+        enriched_location_tag_and_subfields(record) => {tag:, location_code_sf:, call_num_sf:}
 
         locations = record.fields(tag).flat_map { |field|
           field.filter_map { |subfield|
             # skip unless subfield code does not match enriched marc tag subfield code
-            next unless subfield.code == subfield_code
+            next unless subfield.code == location_code_sf
 
-            # skip if subfield value is 'web'
-            # we don't facet for 'web' which is the 'Penn Library Web' location used in Voyager.
-            # this location should eventually go away completely with data cleanup in Alma.
-            next if subfield.value == 'web'
+            location_code = subfield.value
 
-            # skip unless subfield value is a key in location_map
-            # sometimes "happening locations" are mistakenly used in holdings records.
-            # that's a data problem that should be fixed.
-            # here, if we encounter a code we can't map, we ignore it, for faceting purposes
-            next unless location_map.key?(subfield.value.to_sym)
+            next if location_code_to_ignore?(location_map, location_code)
 
-            location_map[subfield.value.to_sym][display_value.to_sym]
+            override = specific_location_override(display_value: display_value, location_code: location_code,
+                                                  field: field, call_num_sf: call_num_sf)
+
+            if override.present?
+              override
+            else
+              location_map[location_code.to_sym][display_value.to_sym]
+            end
           }.flatten.compact_blank
         }.uniq
         if record.tags.intersect?([Enriched::Pub::ELEC_INVENTORY_TAG, Enriched::Api::ELEC_INVENTORY_TAG])
           locations << ONLINE_LIBRARY
         end
+
         locations
       end
 
       private
 
-      # Determine enriched marc location tag ('itm' or 'hld') and subfield code, giving priority to using 'itm' tag and
-      # subfield.
+      # Determine enriched marc location tag, location code subfield, and call number subfield,
+      # giving priority to using 'itm', 'AVA', or 'AVE' fields.
       # @param [MARC::Record]
       # @return [Hash<String, String>] containing location tag and subfield code
       # - `:tag` (String): The enriched marc location tag
-      # - `:subfield_code` (String): The relevant subfield code
-      def location_tag_and_subfield_code(record)
+      # - `:location_code_sf` (String): The subfield code where location code is stored
+      # - `:call_num_sf` (String): The subfield code where call number is stored
+      def enriched_location_tag_and_subfields(record)
         # in holdings records, the shelving location is always the permanent location.
         # in item records, the current location takes into account
         # temporary locations and permanent locations. if you update the item's perm location,
@@ -89,18 +92,57 @@ module PennMARC
         # if the record has an enriched item field present, use it
         if field_defined?(record, Enriched::Pub::ITEM_TAG)
           tag = Enriched::Pub::ITEM_TAG
-          subfield_code = Enriched::Pub::ITEM_CURRENT_LOCATION
-        # if the record has API inventory tags, use them
+          location_code_sf = Enriched::Pub::ITEM_CURRENT_LOCATION
+          call_num_sf = Enriched::Pub::ITEM_CALL_NUMBER
+          # if the record has API inventory tags, use them
         elsif field_defined?(record, Enriched::Api::PHYS_INVENTORY_TAG)
           tag = Enriched::Api::PHYS_INVENTORY_TAG
-          subfield_code = Enriched::Api::PHYS_LOCATION_CODE
-        # otherwise use Pub holding tags
+          location_code_sf = Enriched::Api::PHYS_LOCATION_CODE
+          call_num_sf = Enriched::Api::PHYS_CALL_NUMBER
+          # otherwise use Pub holding tags
         else
           tag = Enriched::Pub::PHYS_INVENTORY_TAG
-          subfield_code = Enriched::Pub::PHYS_LOCATION_CODE
+          location_code_sf = Enriched::Pub::PHYS_LOCATION_CODE
+          call_num_sf = Enriched::Pub::HOLDING_CLASSIFICATION_PART
+        end
+
+        { tag: tag, location_code_sf: location_code_sf, call_num_sf: call_num_sf }
+      end
+
+      # Determines whether to ignore a location code.
+      # We ignore location codes that are not keys in the location map. Sometimes "happening locations" are
+      # mistakenly used in holdings records. That's a data problem that should be fixed. If we encounter a code we can't
+      # map, we ignore it, for faceting purposes. We also ignore the location code 'web'. We don't facet for 'web'
+      # which is the 'Penn Library Web' location used in Voyager. This location should eventually go away completely
+      # with data cleanup in Alma.
+      # @param [location_map] location_map hash with location_code as key and location hash as value
+      # @param [location_code] location_code retrieved from record
+      # @return [Boolean]
+      def location_code_to_ignore?(location_map, location_code)
+        location_map.key?(location_code.to_sym) == false || location_code == WEB_LOCATION_CODE
+      end
+
+      # Retrieves a specific location override based on location code and call number. Specific location overrides are
+      # located in `location_overrides.yml`.
+      # @param [String] display_value
+      # @param [String] location_code
+      # @param [MARC::Field] field
+      # @param [String] call_num_sf
+      # @return [String, Nil]
+      def specific_location_override(display_value:, location_code:, field:, call_num_sf:)
+        return unless display_value == :specific_location
+
+        locations_overrides = Mappers.location_overrides
+
+        call_numbers = subfield_values(field, call_num_sf)
+
+        override = locations_overrides.select do |_key, value|
+          value[:location_code] == location_code && call_numbers.any? { |num| num.match?(value[:call_num_pattern]) }
         end
 
-        { tag: tag, subfield_code: subfield_code }
+        return if override.blank?
+
+        override[override.keys.first][:specific_location]
       end
     end
   end

data/lib/pennmarc/helpers/note.rb

@@ -166,12 +166,19 @@ module PennMARC
         system_details_notes.uniq
       end
 
+      # Retrieve "With" notes for display from field {https://www.loc.gov/marc/bibliographic/bd501.html 501}
+      # @param [Marc::Record] record
+      # @return [Array<String>]
+      def bound_with_show(record)
+        record.fields('501').filter_map { |field| join_subfields(field, &subfield_in?(['a'])).presence }.uniq
+      end
+
       private
 
       # For system details: extract subfield ǂ3 plus other subfields as specified by passed-in block. Pays special
       # attention to punctuation, joining subfield ǂ3 values with a colon-space (': ').
       # @param [MARC::DataField] field
-      # @param [Proc] selector
+      # @param [Proc] &
       # @return [String]
       def sub3_and_other_subs(field, &)
         sub3 = field.filter_map { |sf| trim_trailing('period', sf.value) if sf.code == '3' }.join(': ')

data/lib/pennmarc/helpers/subject.rb

@@ -160,7 +160,6 @@ module PennMARC
 
       # Format a term hash as a string for display
       #
-      # @todo confirm punctuation handling
       # @todo support search field formatting?
       # @param [Symbol] type
       # @param [Hash] term components and information as a hash
@@ -171,11 +170,12 @@ module PennMARC
         # attempt to handle poorly coded record
         normalize_single_subfield(term[:parts].first) if term[:count] == 1 && term[:parts].first.present?
 
-        case type.to_sym
+        case type
         when :facet
-          term[:parts].join('--').strip
+          trim_trailing(:period, term[:parts].join('--').strip)
         when :display
-          "#{term[:parts].join('--')} #{term[:append].join(' ')}".strip
+          display_value = "#{term[:parts].join('--')} #{term[:append].join(' ')}".strip
+          display_value.ends_with?('.') ? display_value : "#{display_value}."
         end
       end
 
@@ -225,9 +225,8 @@ module PennMARC
       #       because we're using (where? - MK) "subdivision count" as a heuristic for the quality level of the
       #       heading. - MG
       # @todo do i need all this?
-      # @todo do i need to handle punctuation? see append_new_part
       # @param [MARC::DataField] field
-      # @return [Hash{Symbol => Integer, Array}, Nil]
+      # @return [Hash{Symbol => Integer, Array, Boolean}, Nil]
       def build_subject_hash(field)
         term_info = { count: 0, parts: [], append: [], uri: nil,
                       local: field.indicator2 == '4' || field.tag.starts_with?('69'), # local subject heading
@@ -243,6 +242,9 @@ module PennMARC
             # filter out PRO/CHR entirely (but only need to check on local heading types)
             return nil if term_info[:local] && subfield.value =~ /^%?(PRO|CHR)([ $]|$)/
 
+            # remove trailing punctuation of previous subject part
+            trim_trailing_commas_or_periods!(term_info[:parts].last)
+
             term_info[:parts] << subfield.value.strip
             term_info[:count] += 1
           when '2'
@@ -252,11 +254,16 @@ module PennMARC
             term_info[:append] << subfield.value.strip # TODO: map relator code?
           when 'b', 'c', 'd', 'p', 'q', 't'
             # these are appended to the last component (part) if possible (i.e., when joined, should have no delimiter)
-            term_info[:parts].last << ", #{subfield.value.strip}"
+            # assume that there is an 'a' preceding value
+            term_info[:parts].last << " #{subfield.value.strip}"
             term_info[:count] += 1
           else
             # the usual case; add a new component to `parts`
             # this typically includes g, v, x, y, z, 4
+
+            # remove trailing punctuation of previous subject part
+            trim_trailing_commas_or_periods!(term_info[:parts].last)
+
             term_info[:parts] << subfield.value.strip
             term_info[:count] += 1
           end
@@ -299,6 +306,15 @@ module PennMARC
         first_part.gsub!(/([[[:alpha:]])])\s+-\s+([[:upper:]]|[[:digit:]]{2,})/, '\1--\2')
         first_part.gsub!(/([[[:alnum:]])])\s+-\s+([[:upper:]])/, '\1--\2')
       end
+
+      # removes trailing comma or period, manipulating the string in place
+      # @param [String, NilClass] subject_part
+      # @return [String, NilClass]
+      def trim_trailing_commas_or_periods!(subject_part)
+        return if subject_part.blank?
+
+        trim_trailing!(:comma, subject_part) || trim_trailing!(:period, subject_part)
+      end
     end
   end
 end

data/lib/pennmarc/mappers.rb

@@ -18,6 +18,11 @@ module PennMARC
         @location ||= load_map('locations.yml')
       end
 
+      # @return [Hash]
+      def location_overrides
+        @location_overrides ||= load_map('location_overrides.yml')
+      end
+
       # @return [Hash]
       def relator
         @relator ||= load_map('relator.yml')

data/lib/pennmarc/mappings/location_overrides.yml

@@ -0,0 +1,5 @@
+albrecht:
+  location_code: 'vanp'
+  call_num_pattern: '^M.*'
+  specific_location: 'Van Pelt - Albrecht Music Library'
+