gdor-indexer 0.3.0 → 0.4.0

checksums.yaml

@@ -1,7 +1,7 @@
 ---
 SHA1:
-  metadata.gz: f5512304eceff43f6099aa3de8864b61d587a6cc
-  data.tar.gz: 3e9e662af728c9d6de0291bff59e9e1cf41b2173
+  metadata.gz: b7d3fef57a3dcc4df36487bead9f96a691ae385d
+  data.tar.gz: f724d1e3bfc0aee77798a10feb3c06939b29c30a
 SHA512:
-  metadata.gz: 0caa4dada6bbde97f147b3630683ed6b689309fc78584ca1aba47eea7421980e7a0ac6835ab55cef50398e8b03d6eca4d480601ffe2ba567c2e399e8f9aa27d5
-  data.tar.gz: 2b3e81c2b3693f4722359ca80efbece1e15b05cd14226986ef1c6bc6578eadcea42a33f1ccb859281c035e3836b0eeed4c8948eeb1444e3f8530f09d910a1b38
+  metadata.gz: bbed1557cd19168a20ca5c26656186206ef80ea83817715fea27fb79218f45ece95f24186f9eeb4fdb17c87b2ccf147556398dcba2fe5bed02db6983951dbe7d
+  data.tar.gz: 578889a4be5005e8a6504313e46b082685a393937a46cd4fe4315756b8da1191ef7a383ec122b5586f6da6ab76ba3157ee325fedf9c87cf67edb24d1cacf0e67

data/.rubocop.yml

@@ -3,6 +3,4 @@ require: rubocop-rspec
 inherit_from: .rubocop_todo.yml
 
 Style/StringLiterals:
-  Enabled: true
-  EnforcedStyle: single_quotes
-  
+  Enabled: false

data/.rubocop_todo.yml

@@ -1,6 +1,6 @@
 # This configuration was generated by
 # `rubocop --auto-gen-config`
-# on 2015-10-28 13:42:33 -0700 using RuboCop version 0.34.2.
+# on 2016-01-06 15:25:10 -0800 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
@@ -17,34 +17,27 @@ Lint/AmbiguousRegexpLiteral:
 Lint/EndAlignment:
   Enabled: false
 
-# Offense count: 6
+# Offense count: 4
 Lint/UselessAssignment:
   Exclude:
     - 'lib/gdor/indexer.rb'
-    - 'spec/unit/gdor_mods_fields_spec.rb'
     - 'spec/unit/indexer_spec.rb'
     - 'spec/unit/solr_doc_builder_spec.rb'
 
-# Offense count: 1
-Lint/Void:
-  Exclude:
-    - 'spec/unit/gdor_mods_fields_spec.rb'
-
-# Offense count: 16
+# Offense count: 21
 Metrics/AbcSize:
-  Max: 82
+  Max: 108
 
 # Offense count: 1
 # Configuration parameters: CountComments.
 Metrics/ClassLength:
-  Exclude:
-    - 'lib/gdor/indexer.rb'
+  Max: 243
 
 # Offense count: 5
 Metrics/CyclomaticComplexity:
   Max: 9
 
-# Offense count: 304
+# Offense count: 307
 # Configuration parameters: AllowURI, URISchemes.
 Metrics/LineLength:
   Max: 258
@@ -52,26 +45,27 @@ Metrics/LineLength:
 # Offense count: 15
 # Configuration parameters: CountComments.
 Metrics/MethodLength:
-  Max: 45
+  Max: 51
 
 # Offense count: 3
 Metrics/PerceivedComplexity:
   Max: 10
 
-# Offense count: 5
+# Offense count: 7
 # Configuration parameters: CustomTransform.
 RSpec/FilePath:
   Exclude:
-    - 'spec/unit/gdor_mods_fields_spec.rb'
     - 'spec/unit/indexer_spec.rb'
+    - 'spec/unit/mods_fields_spec.rb'
+    - 'spec/unit/mods_pub_fields_spec.rb'
+    - 'spec/unit/mods_subject_fields_spec.rb'
     - 'spec/unit/public_xml_fields_spec.rb'
     - 'spec/unit/solr_doc_builder_spec.rb'
     - 'spec/unit/solr_doc_hash_spec.rb'
 
-# Offense count: 359
+# Offense count: 193
 RSpec/InstanceVariable:
   Exclude:
-    - 'spec/unit/gdor_mods_fields_spec.rb'
     - 'spec/unit/indexer_spec.rb'
     - 'spec/unit/public_xml_fields_spec.rb'
     - 'spec/unit/solr_doc_builder_spec.rb'
@@ -106,12 +100,25 @@ Style/DoubleNegation:
   Exclude:
     - 'lib/gdor/indexer/solr_doc_hash.rb'
 
+# Offense count: 1
+# Cop supports --auto-correct.
+# Configuration parameters: EnforcedStyle, SupportedStyles.
+Style/EmptyLinesAroundBlockBody:
+  Exclude:
+    - 'spec/unit/mods_pub_fields_spec.rb'
+
 # Offense count: 3
 # Configuration parameters: MinBodyLength.
 Style/GuardClause:
   Exclude:
     - 'lib/gdor/indexer.rb'
 
+# Offense count: 12
+# Cop supports --auto-correct.
+# Configuration parameters: EnforcedStyle, SupportedStyles.
+Style/MultilineOperationIndentation:
+  Enabled: false
+
 # Offense count: 1
 # Configuration parameters: NamePrefix, NamePrefixBlacklist.
 Style/PredicateName:
@@ -123,8 +130,8 @@ Style/PredicateName:
 # Configuration parameters: EnforcedStyle, SupportedStyles, AllowInnerSlashes.
 Style/RegexpLiteral:
   Exclude:
-    - 'spec/unit/gdor_mods_fields_spec.rb'
     - 'spec/unit/indexer_spec.rb'
+    - 'spec/unit/mods_subject_fields_spec.rb'
 
 # Offense count: 1
 Style/UnlessElse:

data/Gemfile

@@ -1,4 +1,4 @@
 source 'https://rubygems.org'
 
-# Specify your gem's dependencies in spotlight-dor-resources.gemspec
+# Specify your gem's dependencies in gdor-indexer.gemspec
 gemspec

data/gdor-indexer.gemspec

@@ -18,7 +18,7 @@ Gem::Specification.new do |spec|
   spec.require_paths = ['lib']
 
   spec.add_dependency 'harvestdor-indexer'
-  spec.add_dependency 'stanford-mods'
+  spec.add_dependency 'stanford-mods', '>= 1.3.4' # for new pub date methods
   spec.add_dependency 'nokogiri'
   spec.add_dependency 'rsolr'
   spec.add_dependency 'activesupport'

data/lib/gdor/indexer/mods_fields.rb

@@ -36,7 +36,6 @@ module GDor::Indexer::ModsFields
       era_facet: smods_rec.era_facet,
 
       format_main_ssim: format_main_ssim,
-      format: format, # for backwards compatibility
 
       language: smods_rec.sw_language_facet,
       physical: smods_rec.term_values([:physical_description, :extent]),
@@ -46,36 +45,38 @@ module GDor::Indexer::ModsFields
 
       # publication fields
       pub_search: smods_rec.place,
-      pub_date_sort: smods_rec.pub_date_sort,
-      imprint_display: smods_rec.pub_date_display,
+      pub_date_sort: smods_rec.pub_date_sortable_string(false), # include approx dates
+      # these are for single value facet display (in leiu of date slider (pub_year_tisim) and deprecated pub_date)
+      pub_year_no_approx_isi: smods_rec.pub_date_facet_single_value(true),
+      pub_year_w_approx_isi: smods_rec.pub_date_facet_single_value(false),
+      # TODO:  remove pub_date after reindexing existing colls;  deprecated in favor of pub_year_xxx_approx_isi ...
       pub_date: smods_rec.pub_date_facet,
-      pub_date_display: smods_rec.pub_date_display, # pub_date_display may be deprecated
+      # display fields
+      imprint_display: smods_rec.pub_date_display,
+      # pub_date_best_sort_str_value is protected ...
+      creation_year_isi: smods_rec.send(:pub_date_best_sort_str_value, smods_rec.date_created_elements(false)),
+      publication_year_isi: smods_rec.send(:pub_date_best_sort_str_value, smods_rec.date_issued_elements(false)),
 
       all_search: smods_rec.text.gsub(/\s+/, ' ')
     }
 
-    # more pub date fields
-    pub_date_sort = doc_hash[:pub_date_sort]
-    if is_positive_int? pub_date_sort
-      doc_hash[:pub_year_tisim] = pub_date_sort # for date slider
-      # put the displayable year in the correct field, :creation_year_isi for example
-      doc_hash[date_type_sym] = smods_rec.pub_date_sort if date_type_sym
+    # more pub date field processing
+    pub_date_sort_val = doc_hash[:pub_date_sort]
+    if is_positive_int? pub_date_sort_val
+      doc_hash[:pub_year_tisim] = pub_date_sort_val # for date slider
+      # remove leading zeros
+      doc_hash[:creation_year_isi] = remove_leading_zeros(doc_hash[:creation_year_isi]) if doc_hash[:creation_year_isi]
+      doc_hash[:publication_year_isi] = remove_leading_zeros(doc_hash[:publication_year_isi]) if doc_hash[:publication_year_isi]
+    else
+      # turn B.C. into -yyy for display fields
+      doc_hash[:creation_year_isi] = '-' + (1000 + doc_hash[:creation_year_isi].to_i).to_s if doc_hash[:creation_year_isi]
+      doc_hash[:publication_year_isi] = '-' + (1000 + doc_hash[:publication_year_isi].to_i).to_s if doc_hash[:publication_year_isi]
     end
 
     doc_hash
   end
 
-  # select one or more format values from the controlled vocabulary here:
-  #   http://searchworks-solr-lb.stanford.edu:8983/solr/select?facet.field=format&rows=0&facet.sort=index
-  # via stanford-mods gem
-  # @return [Array<String>] value(s) in the SearchWorks controlled vocabulary, or []
-  def format
-    vals = smods_rec.format
-    if vals.empty?
-      logger.warn "#{druid} has no SearchWorks format from MODS - check <typeOfResource> and other implicated MODS elements"
-    end
-    vals
-  end
+  private
 
   # call stanford-mods format_main to get results
   # @return [Array<String>] value(s) in the SearchWorks controlled vocabulary, or []
@@ -87,14 +88,6 @@ module GDor::Indexer::ModsFields
     vals
   end
 
-  # call stanford-mods sw_genre to get results
-  # @return [Array<String>] value(s)
-  def genre_ssim
-    smods_rec.sw_genre
-  end
-
-  protected
-
   # @return true if the string parses into an int, and if so, the int is >= 0
   def is_positive_int?(str)
     str.to_i >= 0
@@ -102,13 +95,9 @@ module GDor::Indexer::ModsFields
     false
   end
 
-  # determines particular flavor of displayable publication year field
-  # @return Solr field name as a symbol
-  def date_type_sym
-    vals = smods_rec.term_values([:origin_info, :dateIssued])
-    return :publication_year_isi if vals && vals.length > 0
-    vals = smods_rec.term_values([:origin_info, :dateCreated])
-    return :creation_year_isi if vals && vals.length > 0
-    nil
+  def remove_leading_zeros(year_sort_str)
+    return '0' if year_sort_str == '0000'
+    return year_sort_str.sub(/^[0:]*/, '') if year_sort_str.start_with?('0')
+    year_sort_str
   end
 end

data/lib/gdor/indexer/version.rb

@@ -1,5 +1,5 @@
 module GDor
   class Indexer
-    VERSION = '0.3.0'
+    VERSION = '0.4.0'
   end
 end

data/spec/unit/mods_fields_spec.rb

@@ -0,0 +1,318 @@
+require 'spec_helper'
+
+describe GDor::Indexer::ModsFields do
+  let(:fake_druid) { 'oo000oo0000' }
+  let(:ns_decl) { "xmlns='#{Mods::MODS_NS}'" }
+  let(:mods_xml) { "<mods #{ns_decl}><note>gdor_mods_fields testing</note></mods>" }
+
+  def sdb_for_mods(m)
+    resource = Harvestdor::Indexer::Resource.new(double, fake_druid)
+    allow(resource).to receive(:public_xml).and_return(nil)
+    allow(resource).to receive(:mods).and_return(Nokogiri::XML(m))
+    i = Harvestdor::Indexer.new
+    i.logger.level = Logger::WARN
+    allow(resource).to receive(:indexer).and_return(i)
+    lgr = Logger.new(StringIO.new)
+    lgr.level = Logger::WARN
+    GDor::Indexer::SolrDocBuilder.new(resource, lgr)
+  end
+
+  # see https://consul.stanford.edu/display/NGDE/Required+and+Recommended+Solr+Fields+for+SearchWorks+documents
+
+  context 'summary_search solr field from <abstract>' do
+    it 'is populated when the MODS has a top level <abstract> element' do
+      m = "<mods #{ns_decl}><abstract>blah blah</abstract></mods>"
+      sdb = sdb_for_mods(m)
+      expect(sdb.doc_hash_from_mods[:summary_search]).to match_array ['blah blah']
+    end
+    it 'has a value for each abstract element' do
+      m = "<mods #{ns_decl}>
+        <abstract>one</abstract>
+        <abstract>two</abstract>
+      </mods>"
+      sdb = sdb_for_mods(m)
+      expect(sdb.doc_hash_from_mods[:summary_search]).to match_array %w(one two)
+    end
+    it 'does not be present when there is no top level <abstract> element' do
+      m = "<mods #{ns_decl}><relatedItem><abstract>blah blah</abstract></relatedItem></mods>"
+      sdb = sdb_for_mods(m)
+      expect(sdb.doc_hash_from_mods[:summary_search]).to be_nil
+    end
+    it 'does not be present if there are only empty abstract elements in the MODS' do
+      m = "<mods #{ns_decl}><abstract/><note>notit</note></mods>"
+      sdb = sdb_for_mods(m)
+      expect(sdb.doc_hash_from_mods[:summary_search]).to be_nil
+    end
+    it 'summary_display should not be populated - it is a copy field' do
+      m = "<mods #{ns_decl}><abstract>blah blah</abstract></mods>"
+      sdb = sdb_for_mods(m)
+      expect(sdb.doc_hash_from_mods[:summary_display]).to be_nil
+    end
+  end # summary_search / <abstract>
+
+  it 'language: should call sw_language_facet in stanford-mods gem to populate language field' do
+    sdb = sdb_for_mods(mods_xml)
+    smr = sdb.smods_rec
+    expect(smr).to receive(:sw_language_facet)
+    sdb.doc_hash_from_mods
+  end
+
+  context 'physical solr field from <physicalDescription><extent>' do
+    it 'is populated when the MODS has mods/physicalDescription/extent element' do
+      m = "<mods #{ns_decl}><physicalDescription><extent>blah blah</extent></physicalDescription></mods>"
+      sdb = sdb_for_mods(m)
+      expect(sdb.doc_hash_from_mods[:physical]).to match_array ['blah blah']
+    end
+    it 'has a value for each extent element' do
+      m = "<mods #{ns_decl}>
+        <physicalDescription>
+          <extent>one</extent>
+          <extent>two</extent>
+        </physicalDescription>
+        <physicalDescription><extent>three</extent></physicalDescription>
+      </mods>"
+      sdb = sdb_for_mods(m)
+      expect(sdb.doc_hash_from_mods[:physical]).to match_array %w(one two three)
+    end
+    it 'does not be present when there is no top level <physicalDescription> element' do
+      m = "<mods #{ns_decl}><relatedItem><physicalDescription><extent>foo</extent></physicalDescription></relatedItem></mods>"
+      sdb = sdb_for_mods(m)
+      expect(sdb.doc_hash_from_mods[:physical]).to be_nil
+    end
+    it 'does not be present if there are only empty physicalDescription or extent elements in the MODS' do
+      m = "<mods #{ns_decl}><physicalDescription/><physicalDescription><extent/></physicalDescription><note>notit</note></mods>"
+      sdb = sdb_for_mods(m)
+      expect(sdb.doc_hash_from_mods[:physical]).to be_nil
+    end
+  end # physical field from physicalDescription/extent
+
+  context 'url_suppl solr field from /mods/relatedItem/location/url' do
+    it 'is populated when the MODS has mods/relatedItem/location/url' do
+      m = "<mods #{ns_decl}><relatedItem><location><url>url.org</url></location></relatedItem></mods>"
+      sdb = sdb_for_mods(m)
+      expect(sdb.doc_hash_from_mods[:url_suppl]).to match_array ['url.org']
+    end
+    it 'has a value for each mods/relatedItem/location/url element' do
+      m = "<mods #{ns_decl}>
+        <relatedItem>
+          <location><url>one</url></location>
+          <location>
+            <url>two</url>
+            <url>three</url>
+          </location>
+        </relatedItem>
+        <relatedItem><location><url>four</url></location></relatedItem>
+      </mods>"
+      sdb = sdb_for_mods(m)
+      expect(sdb.doc_hash_from_mods[:url_suppl]).to match_array %w(one two three four)
+    end
+    it 'does not be populated from /mods/location/url element' do
+      m = "<mods #{ns_decl}><location><url>hi</url></location></mods>"
+      sdb = sdb_for_mods(m)
+      expect(sdb.doc_hash_from_mods[:url_suppl]).to be_nil
+    end
+    it 'does not be present if there are only empty relatedItem/location/url elements in the MODS' do
+      m = "<mods #{ns_decl}>
+        <relatedItem><location><url/></location></relatedItem>
+        <relatedItem><location/></relatedItem>
+        <relatedItem/><note>notit</note></mods>"
+      sdb = sdb_for_mods(m)
+      expect(sdb.doc_hash_from_mods[:url_suppl]).to be_nil
+    end
+  end
+
+  context 'toc_search solr field from <tableOfContents>' do
+    it 'has a value for each tableOfContents element' do
+      m = "<mods #{ns_decl}>
+      <tableOfContents>one</tableOfContents>
+      <tableOfContents>two</tableOfContents>
+      </mods>"
+      sdb = sdb_for_mods(m)
+      expect(sdb.doc_hash_from_mods[:toc_search]).to match_array %w(one two)
+    end
+    it 'does not be present when there is no top level <tableOfContents> element' do
+      m = "<mods #{ns_decl}><relatedItem><tableOfContents>foo</tableOfContents></relatedItem></mods>"
+      sdb = sdb_for_mods(m)
+      expect(sdb.doc_hash_from_mods[:toc_search]).to be_nil
+    end
+    it 'does not be present if there are only empty tableOfContents elements in the MODS' do
+      m = "<mods #{ns_decl}><tableOfContents/><note>notit</note></mods>"
+      sdb = sdb_for_mods(m)
+      expect(sdb.doc_hash_from_mods[:toc_search]).to be_nil
+    end
+  end
+
+  context '#format_main_ssim' do
+    it 'doc_hash_from_mods calls #format_main_ssim' do
+      m = "<mods #{ns_decl}><note>nope</typeOfResource></mods>"
+      sdb = sdb_for_mods(m)
+      expect(sdb).to receive(:format_main_ssim)
+      sdb.doc_hash_from_mods[:format_main_ssim]
+    end
+    it '#format_main_ssim calls stanford-mods.format_main' do
+      m = "<mods #{ns_decl}><note>nope</typeOfResource></mods>"
+      sdb = sdb_for_mods(m)
+      expect(sdb.smods_rec).to receive(:format_main).and_return([])
+      sdb.send(:format_main_ssim)
+    end
+    it 'has a value when MODS data provides' do
+      m = "<mods #{ns_decl}><typeOfResource>still image</typeOfResouce></mods>"
+      sdb = sdb_for_mods(m)
+      expect(sdb.send(:format_main_ssim)).to match_array ['Image']
+    end
+    it 'returns empty Array and logs warning if there is no value' do
+      sdb = sdb_for_mods(mods_xml)
+      expect(sdb.logger).to receive(:warn).with("#{fake_druid} has no SearchWorks Resource Type from MODS - check <typeOfResource> and other implicated MODS elements")
+      expect(sdb.send(:format_main_ssim)).to eq([])
+    end
+  end
+
+  context 'title fields' do
+    let(:title_mods) do
+      "<mods #{ns_decl}>
+        <titleInfo>
+          <title>Jerk</title>
+          <nonSort>The</nonSort>
+          <subTitle>is whom?</subTitle>
+        </titleInfo>
+        <titleInfo>
+          <title>Joke</title>
+        </titleInfo>
+        <titleInfo type='alternative'>
+          <title>Alternative</title>
+        </titleInfo>
+      </mods>"
+    end
+    let(:sdb) { sdb_for_mods(title_mods) }
+    let(:title_doc_hash) { sdb.doc_hash_from_mods }
+
+    it 'calls the appropriate methods in the stanford-mods gem to populate the fields' do
+      smr = sdb.smods_rec
+      expect(smr).to receive(:sw_short_title).at_least(:once)
+      expect(smr).to receive(:sw_full_title).at_least(:once)
+      expect(smr).to receive(:sw_title_display)
+      expect(smr).to receive(:sw_addl_titles)
+      expect(smr).to receive(:sw_sort_title)
+      sdb.doc_hash_from_mods
+    end
+    context 'search fields' do
+      it 'title_245a_search' do
+        expect(title_doc_hash[:title_245a_search]).to eq('The Jerk')
+      end
+      it 'title_245_search' do
+        expect(title_doc_hash[:title_245_search]).to eq('The Jerk : is whom?')
+      end
+      it 'title_variant_search' do
+        expect(title_doc_hash[:title_variant_search]).to match_array %w(Joke Alternative)
+      end
+      it 'title_related_search should not be populated from MODS' do
+        expect(title_doc_hash[:title_related_search]).to be_nil
+      end
+    end
+    context 'display fields' do
+      it 'title_display' do
+        expect(title_doc_hash[:title_display]).to eq('The Jerk : is whom?')
+      end
+      it 'title_245a_display' do
+        expect(title_doc_hash[:title_245a_display]).to eq('The Jerk')
+      end
+      it 'title_245c_display should not be populated from MODS' do
+        expect(title_doc_hash[:title_245c_display]).to be_nil
+      end
+      it 'title_full_display' do
+        expect(title_doc_hash[:title_full_display]).to eq('The Jerk : is whom?')
+      end
+      it 'removes trailing commas in title_display' do
+        title_mods = "<mods #{ns_decl}>
+        <titleInfo><title>Jerk</title><nonSort>The</nonSort><subTitle>is whom,</subTitle></titleInfo>
+        <titleInfo><title>Joke</title></titleInfo>
+        <titleInfo type='alternative'><title>Alternative</title></titleInfo>
+        </mods>"
+        sdb = sdb_for_mods(title_mods)
+        title_doc_hash = sdb.doc_hash_from_mods
+        expect(title_doc_hash[:title_display]).to eq('The Jerk : is whom')
+      end
+      it 'title_variant_display should not be populated - it is a copy field' do
+        expect(title_doc_hash[:title_variant_display]).to be_nil
+      end
+    end
+    it 'title_sort' do
+      expect(title_doc_hash[:title_sort]).to eq('Jerk is whom')
+    end
+  end # title fields
+
+  context 'author fields' do
+    let(:name_mods) do
+      "<mods #{ns_decl}>
+        <name type='personal'>
+          <namePart type='given'>John</namePart>
+          <namePart type='family'>Huston</namePart>
+          <role><roleTerm type='code' authority='marcrelator'>drt</roleTerm></role>
+          <displayForm>q</displayForm>
+        </name>
+        <name type='personal'><namePart>Crusty The Clown</namePart></name>
+        <name type='corporate'><namePart>Watchful Eye</namePart></name>
+        <name type='corporate'>
+          <namePart>Exciting Prints</namePart>
+          <role><roleTerm type='text'>lithographer</roleTerm></role>
+        </name>
+        <name type='conference'><namePart>conference</namePart></name>
+      </mods>"
+    end
+    let(:sdb) { sdb_for_mods(name_mods) }
+    let(:author_doc_hash) { sdb.doc_hash_from_mods }
+
+    it 'calls the appropriate methods in the stanford-mods gem to populate the fields' do
+      smr = sdb.smods_rec
+      expect(smr).to receive(:sw_main_author)
+      expect(smr).to receive(:sw_addl_authors)
+      expect(smr).to receive(:sw_person_authors).exactly(3).times
+      expect(smr).to receive(:sw_impersonal_authors)
+      expect(smr).to receive(:sw_corporate_authors)
+      expect(smr).to receive(:sw_meeting_authors)
+      expect(smr).to receive(:sw_sort_author)
+      sdb.doc_hash_from_mods
+    end
+    context 'search fields' do
+      it 'author_1xx_search' do
+        expect(author_doc_hash[:author_1xx_search]).to eq('Crusty The Clown')
+      end
+      it 'author_7xx_search' do
+        skip 'Should this return all authors? or only 7xx authors?'
+        expect(author_doc_hash[:author_7xx_search]).to match_array ['q', 'Watchful Eye', 'Exciting Prints', 'conference']
+      end
+      it 'author_8xx_search should not be populated from MODS' do
+        expect(author_doc_hash[:author_8xx_search]).to be_nil
+      end
+    end
+    context 'facet fields' do
+      it 'author_person_facet' do
+        expect(author_doc_hash[:author_person_facet]).to match_array ['q', 'Crusty The Clown']
+      end
+      it 'author_other_facet' do
+        expect(author_doc_hash[:author_other_facet]).to match_array ['Watchful Eye', 'Exciting Prints', 'conference']
+      end
+    end
+    context 'display fields' do
+      it 'author_person_display' do
+        expect(author_doc_hash[:author_person_display]).to match_array ['q', 'Crusty The Clown']
+      end
+      it 'author_person_full_display' do
+        expect(author_doc_hash[:author_person_full_display]).to match_array ['q', 'Crusty The Clown']
+      end
+      it 'author_corp_display' do
+        expect(author_doc_hash[:author_corp_display]).to match_array ['Watchful Eye', 'Exciting Prints']
+      end
+      it 'author_meeting_display' do
+        expect(author_doc_hash[:author_meeting_display]).to match_array ['conference']
+      end
+    end
+    it 'author_sort' do
+      expect(author_doc_hash[:author_sort]).to eq('Crusty The Clown')
+    end
+  end # author fields
+
+  # subject fields moved to mods_subject_fields_spec.rb
+
+  # publication fields moved to mods_pub_fields_spec.rb
+end