spotlight-dor-resources 0.0.3 → 0.0.4

checksums.yaml

@@ -1,7 +1,7 @@
 ---
 SHA1:
-  metadata.gz: 5b035c7c9d1aa81f28332f8304294beb43a960ab
-  data.tar.gz: 1658a13bf5ff55aa7f0d67cf7770c89e17648f2d
+  metadata.gz: 8b1e1f0ddb963f81f8fd302a284ef53592146ef9
+  data.tar.gz: 03687ee4c19706fa434993ce33ad59beac3b522a
 SHA512:
-  metadata.gz: fa6d70d3f6530b159e9bec06a3c8b2219bb836a5a3608d291c047ae7d405d653ccf02d0f4b028523a0a942848d1b9cd2070fe42cd28e786496a8059c670c28df
-  data.tar.gz: a99351ec2d515332d76582c6b178be12021834c499d3725225b2cf2727871dd794c800a913dd87a40091cf807b76145489797cddf00ad04b1f7f97209c04ec03
+  metadata.gz: 01970798dc91a7dd7b3f451244e71573d6c1129eb2c59a7f7758b3fdfb4d7d70c3d5bf8e3d99d1a7300b404dbc7142869898b1792729f8ab1099beb1f7113cd1
+  data.tar.gz: 121941bdd33d142d20f27bcacd25d112327324d5e1f84e0f190896bb9fa0c932562cdaeea2a29dae846d84b940085e3bcdd3b92a4f832854e74c24f46269621d

data/.rubocop.yml

@@ -4,9 +4,24 @@ require: rubocop-rspec
 
 AllCops:
   Exclude:
+    - 'Gemfile'
+    - 'bin/**/*'
+    - 'db/**/*'
+    - 'config/**/*'
     - 'spec/internal/**/*'
     - 'spec/test_app_templates/**/*'
+    - 'spec/spec_helper.rb'
+    - 'spec/teaspoon_env.rb'
+    - 'vendor/**/*'
+  RunRailsCops: true
 
-Metrics/ClassLength:
+Metrics/LineLength:
+  Max: 130
+
+Style/StringLiterals:
+  Enabled: true
+  EnforcedStyle: single_quotes
+
+RSpec/DescribeClass:
   Exclude:
-    - 'lib/spotlight/dor/indexer.rb'
+    - 'spec/integration/*'

data/.rubocop_todo.yml

@@ -1,188 +1,3 @@
-# This configuration was generated by
-# `rubocop --auto-gen-config`
-# on 2015-10-29 09:20:45 -0700 using RuboCop version 0.34.2.
-# 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: 1
-# Configuration parameters: CountComments.
-Metrics/ClassLength:
-  Max: 102
-
-# Offense count: 84
-# Configuration parameters: AllowURI, URISchemes.
-Metrics/LineLength:
-  Max: 171
-
-# Offense count: 1
-# Configuration parameters: CountComments.
-Metrics/MethodLength:
-  Max: 13
-
-# Offense count: 1
-# Cop supports --auto-correct.
-Performance/StringReplacement:
-  Exclude:
-    - 'lib/spotlight/dor/indexer.rb'
-
-# Offense count: 2
-RSpec/DescribeClass:
-  Exclude:
-    - 'spec/integration/gdor_integration_spec.rb'
-    - 'spec/integration/indexer_integration_spec.rb'
-
-# Offense count: 4
-RSpec/DescribedClass:
-  Exclude:
-    - 'spec/models/spotlight/resources/purl_spec.rb'
-    - 'spec/models/spotlight/resources/searchworks_spec.rb'
-
-# Offense count: 26
-# Configuration parameters: CustomTransform, IgnoredWords.
-RSpec/ExampleWording:
-  Exclude:
-    - 'spec/integration/gdor_integration_spec.rb'
-    - 'spec/models/spotlight/resources/purl_spec.rb'
-    - 'spec/models/spotlight/resources/searchworks_spec.rb'
-
-# Offense count: 4
-# Cop supports --auto-correct.
-# Configuration parameters: EnforcedStyle, SupportedStyles.
-Style/AndOr:
-  Exclude:
-    - 'lib/spotlight/dor/indexer.rb'
-    - 'spec/spec_helper.rb'
-
-# Offense count: 11
-# Cop supports --auto-correct.
-# Configuration parameters: EnforcedStyle, SupportedStyles.
-Style/BracesAroundHashParameters:
-  Exclude:
-    - 'spec/models/spotlight/resources/purl_spec.rb'
-    - 'spec/models/spotlight/resources/searchworks_spec.rb'
-
-# Offense count: 6
-# Configuration parameters: EnforcedStyle, SupportedStyles.
 Style/ClassAndModuleChildren:
-  Exclude:
-    - 'app/models/spotlight/resources/dor_resource.rb'
-    - 'app/models/spotlight/resources/harvestdor.rb'
-    - 'app/models/spotlight/resources/purl.rb'
-    - 'app/models/spotlight/resources/searchworks.rb'
-    - 'lib/spotlight/dor/indexer.rb'
-    - 'lib/spotlight/dor/resources/engine.rb'
-
-# Offense count: 7
-# Configuration parameters: Exclude.
-Style/Documentation:
-  Exclude:
-    - 'app/models/spotlight/resources/dor_resource.rb'
-    - 'app/models/spotlight/resources/purl.rb'
-    - 'app/models/spotlight/resources/searchworks.rb'
-    - 'lib/spotlight/dor/indexer.rb'
-    - 'lib/spotlight/dor/resources.rb'
-    - 'lib/spotlight/dor/resources/engine.rb'
-    - 'lib/spotlight/dor/resources/version.rb'
-
-# Offense count: 2
-Style/DoubleNegation:
-  Exclude:
-    - 'app/models/spotlight/resources/purl.rb'
-    - 'app/models/spotlight/resources/searchworks.rb'
-
-# Offense count: 1
-# Cop supports --auto-correct.
-Style/EmptyLines:
-  Exclude:
-    - 'spec/spec_helper.rb'
-
-# Offense count: 4
-# Cop supports --auto-correct.
-# Configuration parameters: EnforcedStyle, SupportedStyles.
-Style/EmptyLinesAroundClassBody:
-  Exclude:
-    - 'app/models/spotlight/resources/purl.rb'
-    - 'app/models/spotlight/resources/searchworks.rb'
-    - 'lib/spotlight/dor/resources/engine.rb'
-
-# Offense count: 1
-# Cop supports --auto-correct.
-# Configuration parameters: EnforcedStyle, SupportedStyles.
-Style/EmptyLinesAroundModuleBody:
-  Exclude:
-    - 'lib/spotlight/dor/resources.rb'
-
-# Offense count: 9
-# Cop supports --auto-correct.
-# Configuration parameters: EnforcedStyle, SupportedStyles.
-Style/MethodDefParentheses:
-  Enabled: false
-
-# Offense count: 4
-# Cop supports --auto-correct.
-# Configuration parameters: EnforcedStyle, SupportedStyles, AllowInnerSlashes.
-Style/RegexpLiteral:
-  Exclude:
-    - 'app/models/spotlight/resources/purl.rb'
-    - 'app/models/spotlight/resources/searchworks.rb'
-
-# Offense count: 2
-# Cop supports --auto-correct.
-Style/SpaceAfterComma:
-  Exclude:
-    - 'lib/spotlight/dor/indexer.rb'
-
-# Offense count: 2
-# Cop supports --auto-correct.
-# Configuration parameters: EnforcedStyle, SupportedStyles, EnforcedStyleForEmptyBraces, SpaceBeforeBlockParameters.
-Style/SpaceInsideBlockBraces:
-  Enabled: false
-
-# Offense count: 31
-# Cop supports --auto-correct.
-# Configuration parameters: EnforcedStyle, EnforcedStyleForEmptyBraces, SupportedStyles.
-Style/SpaceInsideHashLiteralBraces:
-  Enabled: false
-
-# Offense count: 122
-# Cop supports --auto-correct.
-# Configuration parameters: EnforcedStyle, SupportedStyles.
-Style/StringLiterals:
-  Enabled: false
-
-# Offense count: 1
-# Cop supports --auto-correct.
-# Configuration parameters: IgnoredMethods.
-Style/SymbolProc:
-  Exclude:
-    - 'lib/spotlight/dor/indexer.rb'
-
-# Offense count: 2
-# Cop supports --auto-correct.
-# Configuration parameters: EnforcedStyle, SupportedStyles.
-Style/TrailingBlankLines:
-  Exclude:
-    - 'spec/models/spotlight/resources/searchworks_spec.rb'
-    - 'spec/spec_helper.rb'
-
-# Offense count: 12
-# Cop supports --auto-correct.
-Style/TrailingWhitespace:
-  Exclude:
-    - 'lib/spotlight/dor/resources.rb'
-    - 'spec/models/spotlight/resources/searchworks_spec.rb'
-
-# Offense count: 1
-# Cop supports --auto-correct.
-# Configuration parameters: ExactNameMatch, AllowPredicates, AllowDSLWriters, IgnoreClassMethods, Whitelist.
-Style/TrivialAccessors:
-  Exclude:
-    - 'lib/spotlight/dor/indexer.rb'
-
-# Offense count: 4
-# Configuration parameters: EnforcedStyle, SupportedStyles.
-Style/VariableName:
-  Enabled: false
+  Enabled: false

data/app/models/spotlight/resources/dor_resource.rb

@@ -1,4 +1,5 @@
 module Spotlight::Resources
+  # Base Resource indexer for objects in DOR
   class DorResource < Spotlight::Resource
     ##
     # Generate solr documents for the DOR resources identified by this object

data/app/models/spotlight/resources/purl.rb

@@ -1,14 +1,14 @@
 module Spotlight::Resources
+  # Resource provider for PURL pages
   class Purl < Spotlight::Resources::DorResource
     self.weight = -1000
 
-    def self.can_provide? res
-      !!(res.url =~ /^https?:\/\/purl.stanford.edu/)
+    def self.can_provide?(res)
+      res.url.match(%r{^https?://purl.stanford.edu/}).present?
     end
 
     def doc_id
-      url.match(/^https?:\/\/purl.stanford.edu\/([^#\/\.]+)/)[1]
+      url.match(%r{^https?://purl.stanford.edu/([^#/\.]+)})[1]
     end
-
   end
 end

data/app/models/spotlight/resources/searchworks.rb

@@ -1,15 +1,14 @@
 module Spotlight::Resources
+  # Resource provider for searchworks pages
   class Searchworks < Spotlight::Resources::DorResource
-
     self.weight = -1000
 
-    def self.can_provide? res
-      !!(res.url =~ /^https?:\/\/searchworks[^\.]*.stanford.edu/)
+    def self.can_provide?(res)
+      res.url.match(%r{^https?://searchworks[^\.]*.stanford.edu/}).present?
     end
 
     def doc_id
-      url.match(/^https?:\/\/searchworks[^\.]*.stanford.edu\/.*view\/([^\/\.#]+)/)[1]
+      url.match(%r{^https?://searchworks[^\.]*.stanford.edu/.*view/([^/\.#]+)})[1]
     end
-
   end
 end

data/lib/spotlight/dor/indexer.rb

@@ -1,171 +1,233 @@
+# rubocop:disable Metrics/ClassLength
 # external gems
 require 'gdor/indexer'
 require 'solrizer'
-# Base class to harvest from DOR via harvestdor gem
+
 module Spotlight::Dor
+  # Base class to harvest from DOR via harvestdor gem
   class Indexer < GDor::Indexer
-    # add contentMetadata fields
-    before_index do |sdb, solr_doc|
-      Solrizer.insert_field(solr_doc, 'content_metadata_type', sdb.public_xml.xpath("/publicObject/contentMetadata/@type").text, :symbol, :displayable)
-
-      sdb.public_xml.xpath("/publicObject/contentMetadata").xpath('resource/file[@mimetype="image/jp2"]').each do |node|
-        file_id = node.attr('id').gsub(".jp2", '')
-
-        if node.attr('id') =~ /jp2$/ and !solr_doc[Solrizer.solr_name('content_metadata_first_image_file_name', :displayable)]
-          Solrizer.insert_field(solr_doc, 'content_metadata_first_image_file_name', file_id, :displayable)
-          Solrizer.insert_field(solr_doc, 'content_metadata_first_image_width', node.xpath('./imageData/@width').text, :displayable)
-          Solrizer.insert_field(solr_doc, 'content_metadata_first_image_height', node.xpath('./imageData/@height').text, :displayable)
-        end
-
-        Solrizer.insert_field(solr_doc, 'content_metadata_image_iiif_info', "https://stacks.stanford.edu/image/iiif/#{solr_doc[:id]}%2F#{file_id}/info.json", :displayable)
-        Solrizer.insert_field(solr_doc, 'thumbnail_square_url', "https://stacks.stanford.edu/image/#{solr_doc[:id]}/#{file_id}_square", :displayable)
-        Solrizer.insert_field(solr_doc, 'thumbnail_url', "https://stacks.stanford.edu/image/#{solr_doc[:id]}/#{file_id}_thumb", :displayable)
-        Solrizer.insert_field(solr_doc, 'large_image_url', "https://stacks.stanford.edu/image/#{solr_doc[:id]}/#{file_id}_large", :displayable)
-        Solrizer.insert_field(solr_doc, 'full_image_url', "https://stacks.stanford.edu/image/#{solr_doc[:id]}/#{file_id}_full", :displayable)
-      end
-    end
-
     # tweak author_sort field from stanford-mods
     before_index do |_sdb, solr_doc|
-      solr_doc[:author_sort] &&= solr_doc[:author_sort].gsub("\uFFFF", "\uFFFD")
+      solr_doc[:author_sort] &&= solr_doc[:author_sort].tr("\uFFFF", "\uFFFD")
     end
 
     # add fields from raw mods
-    before_index :add_box
-    # see comment with add_donor_tags about Feigenbaum specific donor tags data
-    before_index :add_donor_tags
     before_index :add_genre
-    before_index :add_folder
-    before_index :add_folder_name
-    before_index :add_series
-    before_index :mods_cartographics_indexing
 
-    def solr_client
-      @solr_client
-    end
-
-    def solr_document resource
+    def solr_document(resource)
       doc_hash = super
       run_hook :before_index, resource, doc_hash
       doc_hash
     end
 
-    def resource druid
+    def resource(druid)
       Harvestdor::Indexer::Resource.new harvestdor, druid
     end
 
     private
 
-    # add the box number to solr_doc as box_ssi field (note: single valued!)
-    #   data in location/physicalLocation or in relatedItem/location/physicalLocation
-    # TODO:  push this up to stanford-mods gem?  or should it be hierarchical series/box/folder?
-    def add_box(sdb, solr_doc)
-      # see spec for data from actual collections
-      #   _location.physicalLocation should find top level and relatedItem
-      box_num = sdb.smods_rec._location.physicalLocation.map do |node|
-        val = node.text
-        # note that this will also find Flatbox or Flat-box
-        match_data = val.match(/Box ?:? ?([^,|(Folder)]+)/i)
-        match_data[1].strip if match_data.present?
-      end.compact
-
-      solr_doc['box_ssi'] = box_num.first
+    # add plain MODS <genre> element data, not the SearchWorks genre values
+    def add_genre(sdb, solr_doc)
+      insert_field solr_doc, 'genre', sdb.smods_rec.genre.content, :symbol # this is a _ssim field
     end
 
-    # This new donor_tags_sim field was added in October 2015 specifically for the Feigenbaum exhibit.  It is very likely
-    #  it will go ununsed by other projects, but should be benign (since this field will not be created if this specific MODs note is not found.)
-    #  Later refactoring could include exhibit specific fields.   Peter Mangiafico
-    def add_donor_tags sdb, solr_doc
-      donor_tags = sdb.smods_rec.note.select { |n| n.displayLabel == 'Donor tags' }.map(&:content)
-      insert_field solr_doc, 'donor_tags', donor_tags, :symbol # this is a _ssim field
-    end
+    concerning :PhysicalLocation do
+      included do
+        before_index :add_box
+        before_index :add_folder
+        before_index :add_location
+        before_index :add_series
+      end
 
-    # add the folder number to solr_doc as folder_ssi field (note: single valued!)
-    #   data in location/physicalLocation or in relatedItem/location/physicalLocation
-    # TODO:  push this up to stanford-mods gem?  or should it be hierarchical series/box/folder?
-    def add_folder(sdb, solr_doc)
-      # see spec for data from actual collections
-      #   _location.physicalLocation should find top level and relatedItem
-      folder_num = sdb.smods_rec._location.physicalLocation.map do |node|
-        val = node.text
-
-        match_data = if val =~ /\|/
-                       # we assume the data is pipe-delimited, and may contain commas within values
-                       val.match(/Folder ?:? ?([^|]+)/)
-                     else
-                       # the data should be comma-delimited, and may not contain commas within values
-                       val.match(/Folder ?:? ?([^,]+)/)
-                     end
-
-        match_data[1].strip if match_data.present?
-      end.compact
-
-      solr_doc['folder_ssi'] = folder_num.first
-    end
+      # add the box number to solr_doc as box_ssi field (note: single valued!)
+      #   data in location/physicalLocation or in relatedItem/location/physicalLocation
+      # TODO:  push this up to stanford-mods gem?  or should it be hierarchical series/box/folder?
+      def add_box(sdb, solr_doc)
+        # see spec for data from actual collections
+        #   _location.physicalLocation should find top level and relatedItem
+        box_num = sdb.smods_rec._location.physicalLocation.map do |node|
+          val = node.text
+          # note that this will also find Flatbox or Flat-box
+          match_data = val.match(/Box ?:? ?([^,|(Folder)]+)/i)
+          match_data[1].strip if match_data.present?
+        end.compact
+
+        solr_doc['box_ssi'] = box_num.first
+      end
+
+      # add the folder number to solr_doc as folder_ssi field (note: single valued!)
+      #   data in location/physicalLocation or in relatedItem/location/physicalLocation
+      # TODO:  push this up to stanford-mods gem?  or should it be hierarchical series/box/folder?
+      def add_folder(sdb, solr_doc)
+        # see spec for data from actual collections
+        #   _location.physicalLocation should find top level and relatedItem
+        folder_num = sdb.smods_rec._location.physicalLocation.map do |node|
+          val = node.text
+
+          match_data = if val =~ /\|/
+                         # we assume the data is pipe-delimited, and may contain commas within values
+                         val.match(/Folder ?:? ?([^|]+)/)
+                       else
+                         # the data should be comma-delimited, and may not contain commas within values
+                         val.match(/Folder ?:? ?([^,]+)/)
+                       end
+
+          match_data[1].strip if match_data.present?
+        end.compact
+
+        solr_doc['folder_ssi'] = folder_num.first
+      end
 
-    # add the folder name to solr_doc as folder_name_ssi field (note: single valued!)
-    #   data is specific to Feigenbaum collection and is in <note type='preferred citation'>
-    def add_folder_name(sdb, solr_doc)
-      # see spec for data examples
-      preferred_citation = sdb.smods_rec.note.select { |n| n.type_at == 'preferred citation' }.map(&:content)
-      match_data = preferred_citation.first.match(/Title: +(.+)/i) if preferred_citation.present?
-      solr_doc['folder_name_ssi'] = match_data[1].rstrip if match_data.present?
+      # add the physicalLocation as location_ssi field (note: single valued!)
+      #   but only if it has series, box or folder data
+      #   data in location/physicalLocation or in relatedItem/location/physicalLocation
+      # TODO:  push this up to stanford-mods gem?  or should it be hierarchical series/box/folder?
+      def add_location(sdb, solr_doc)
+        # see spec for data from actual collections
+        #   _location.physicalLocation should find top level and relatedItem
+        loc = sdb.smods_rec._location.physicalLocation.map do |node|
+          node.text if node.text.match(/.*(Series)|(Accession)|(Folder)|(Box).*/i)
+        end.compact
+
+        solr_doc['location_ssi'] = loc.first
+      end
+
+      # add the series/accession 'number' to solr_doc as series_ssi field (note: single valued!)
+      #   data in location/physicalLocation or in relatedItem/location/physicalLocation
+      # TODO:  push this up to stanford-mods gem?  or should it be hierarchical series/box/folder?
+      def add_series(sdb, solr_doc)
+        # see spec for data from actual collections
+        #   _location.physicalLocation should find top level and relatedItem
+        series_num = sdb.smods_rec._location.physicalLocation.map do |node|
+          val = node.text
+          # feigenbaum uses 'Accession'
+          match_data = val.match(/(?:(?:Series)|(?:Accession)):? ([^,|]+)/i)
+          match_data[1].strip if match_data.present?
+        end.compact
+
+        solr_doc['series_ssi'] = series_num.first
+      end
     end
 
-    # add plain MODS <genre> element data, not the SearchWorks genre values
-    def add_genre sdb, solr_doc
-      insert_field solr_doc, 'genre', sdb.smods_rec.genre.content, :symbol # this is a _ssim field
+    concerning :ContentMetadata do
+      included do
+        before_index :add_content_metadata_fields
+      end
+
+      def add_content_metadata_fields(sdb, solr_doc)
+        content_metadata = sdb.public_xml.at_xpath('/publicObject/contentMetadata')
+        return unless content_metadata.present?
+
+        Solrizer.insert_field(solr_doc, 'content_metadata_type', content_metadata['type'], :symbol, :displayable)
+
+        images = content_metadata.xpath('resource/file[@mimetype="image/jp2"]').select { |node| node.attr('id') =~ /jp2$/ }
+
+        add_thumbnail_fields(images.first, solr_doc) if images.first
+
+        images.each do |image|
+          add_image_fields(image, solr_doc)
+        end
+      end
+
+      private
+
+      def add_thumbnail_fields(node, solr_doc)
+        file_id = node.attr('id').gsub('.jp2', '')
+        image_data = node.at_xpath('./imageData')
+
+        Solrizer.insert_field(solr_doc, 'content_metadata_first_image_file_name', file_id, :displayable)
+        Solrizer.insert_field(solr_doc, 'content_metadata_first_image_width', image_data['width'], :displayable)
+        Solrizer.insert_field(solr_doc, 'content_metadata_first_image_height', image_data['height'], :displayable)
+      end
+
+      def add_image_fields(node, solr_doc)
+        file_id = node.attr('id').gsub('.jp2', '')
+        base_url = stacks_iiif_url(solr_doc[:id], file_id)
+
+        Solrizer.insert_field(solr_doc, 'content_metadata_image_iiif_info', "#{base_url}/info.json", :displayable)
+        Solrizer.insert_field(solr_doc, 'thumbnail_square_url', "#{base_url}/square/100,100/0/default.jpg", :displayable)
+        Solrizer.insert_field(solr_doc, 'thumbnail_url', "#{base_url}/full/!400,400/0/default.jpg", :displayable)
+        Solrizer.insert_field(solr_doc, 'large_image_url', "#{base_url}/full/pct:25/0/default.jpg", :displayable)
+        Solrizer.insert_field(solr_doc, 'full_image_url', "#{base_url}/full/full/0/default.jpg", :displayable)
+      end
+
+      def stacks_iiif_url(druid, file_name)
+        "#{Spotlight::Dor::Resources::Engine.config.stacks_iiif_url}/#{druid}%2F#{file_name}"
+      end
     end
 
-    # add the series/accession 'number' to solr_doc as series_ssi field (note: single valued!)
-    #   data in location/physicalLocation or in relatedItem/location/physicalLocation
-    # TODO:  push this up to stanford-mods gem?  or should it be hierarchical series/box/folder?
-    def add_series(sdb, solr_doc)
-      # see spec for data from actual collections
-      #   _location.physicalLocation should find top level and relatedItem
-      series_num = sdb.smods_rec._location.physicalLocation.map do |node|
-        val = node.text
-        # feigenbaum uses 'Accession'
-        match_data = val.match(/(?:(?:Series)|(?:Accession)):? ([^,|]+)/i)
-        match_data[1].strip if match_data.present?
-      end.compact
-
-      solr_doc['series_ssi'] = series_num.first
+    concerning :FeigenbaumSpecificFields do
+      # These fields were specifically for the Feigenbaum exhibit.  It is very
+      # likely it will go ununsed by other projects, but should be benign (since this field will not be created if
+      # this specific MODs note is not found.). Future work could refactor this to
+      # only create these fields on an as-needed basis.
+
+      included do
+        before_index :add_donor_tags
+        before_index :add_folder_name
+      end
+
+      def add_donor_tags(sdb, solr_doc)
+        donor_tags = sdb.smods_rec.note.select { |n| n.displayLabel == 'Donor tags' }.map(&:content)
+        insert_field solr_doc, 'donor_tags', donor_tags, :symbol # this is a _ssim field
+      end
+
+      # add the folder name to solr_doc as folder_name_ssi field (note: single valued!)
+      #   data is specific to Feigenbaum collection and is in <note type='preferred citation'>
+      def add_folder_name(sdb, solr_doc)
+        # see spec for data examples
+        preferred_citation = sdb.smods_rec.note.select { |n| n.type_at == 'preferred citation' }.map(&:content)
+        match_data = preferred_citation.first.match(/Title: +(.+)/i) if preferred_citation.present?
+        solr_doc['folder_name_ssi'] = match_data[1].strip if match_data.present?
+      end
     end
 
-    # rubocop:disable Metrics/AbcSize
-    def mods_cartographics_indexing sdb, solr_doc
-      insert_field(solr_doc, "coordinates", Array(sdb.smods_rec.subject.cartographics.coordinates).map { |n| n.text }, :stored_searchable)
+    concerning :CartographicIndexing do
+      included do
+        before_index :mods_cartographics_indexing
+      end
 
-      Array(sdb.smods_rec.subject.cartographics.coordinates).map do |n|
-        next unless n.text =~ /^\(/ and n.text =~ /\)$/
+      def mods_cartographics_indexing(sdb, solr_doc)
+        coordinates = Array(sdb.smods_rec.subject.cartographics.coordinates)
 
-        bbox = n.text.gsub(/[\(\)]/, '')
+        insert_field(solr_doc, 'coordinates', coordinates.map(&:text), :stored_searchable)
 
-        lng, lat = bbox.split('/')
+        solr_doc['point_bbox'] ||= []
+        solr_doc['point_bbox'] += coords_to_bboxes(coordinates)
+      end
+
+      private
 
-        minX,maxX = lng.split('--').map { |x| coord_to_decimal(x) }
-        maxY,minY = lat.split('--').map { |x| coord_to_decimal(x) }
+      def coords_to_bboxes(coordinates)
+        coordinates.select { |n| n.text =~ /^\(.*\)$/ }.map do |n|
+          coord_to_bbox(n.text)
+        end
+      end
 
-        solr_doc["point_bbox"] ||= []
-        solr_doc["point_bbox"] << "#{minX} #{minY} #{maxX} #{maxY}"
+      def coord_to_bbox(coord)
+        bbox = coord.delete('(').delete(')')
+
+        lng, lat = bbox.split('/')
+
+        min_x, max_x = lng.split('--').map { |x| coord_to_decimal(x) }
+        max_y, min_y = lat.split('--').map { |y| coord_to_decimal(y) }
+        "#{min_x} #{min_y} #{max_x} #{max_y}"
       end
-    end
-    # rubocop:enable Metrics/AbcSize
 
-    def coord_to_decimal point
-      regex = /(?<dir>[NESW])\s*(?<deg>\d+)°(?:(?<sec>\d+)ʹ)?/
-      match = regex.match(point)
-      dec = 0
+      def coord_to_decimal(point)
+        regex = /(?<dir>[NESW])\s*(?<deg>\d+)°(?:(?<sec>\d+)ʹ)?/
+        match = regex.match(point)
+        dec = 0
 
-      dec += match['deg'].to_i
-      dec += match['sec'].to_f / 60
-      dec = -1 * dec if match['dir'] == "W" or match['dir'] == "S"
+        dec += match['deg'].to_i
+        dec += match['sec'].to_f / 60
+        dec = -1 * dec if match['dir'] == 'W' || match['dir'] == 'S'
 
-      dec
+        dec
+      end
     end
 
-    def insert_field solr_doc, field, values, *args
+    def insert_field(solr_doc, field, values, *args)
       Array(values).each do |v|
         Solrizer.insert_field solr_doc, field, v, *args
       end