gdor-indexer 0.4.1 → 0.5.0

checksums.yaml

@@ -1,7 +1,7 @@
 ---
 SHA1:
-  metadata.gz: b01625bc9ae63bd07de7edbee92754e2d5c3ce82
-  data.tar.gz: 9586786a5f096aeaf9a7cacac5aa4d07a846ac38
+  metadata.gz: 4724a7e6363dc6d287c4bf1bf6033e000930bf90
+  data.tar.gz: 72fbd1746a114c918ebc9dd838f9674f7fddb8b0
 SHA512:
-  metadata.gz: c82da4f86166c0864a0ad5ed969304b8ee8a357d07efa71738d7a7ce50976cb509568ff812b9ca3b28a44d514237a37ff69eb43a8a617cd28e6aadf2e6794378
-  data.tar.gz: 9bc2a66dbf0652206ca609b5658eebf0d80d32695c0ccf03cace3bf58506650dc212f7b49e09bf6ecfb50c2f6f1c5a5389a913345318e1bf3b212f74b0dcdad9
+  metadata.gz: 88b7024e0faf8b01d0091b6f53e1683d9f9fc9ba6602d4905d2078e682d7cdb284c95e63d362698ef22709961f313f002ec483dcad43f5f4e224149e6269d339
+  data.tar.gz: 8b9b76607dabc0f1409a66b37efb1fd0e07c8afc29a4c36d65a3f144b60b6b531afb3251512a97cd75a47164ce8b599cdf79a0d12fd795e2f85c8385983fbb21

data/.rubocop_todo.yml

@@ -1,6 +1,6 @@
 # This configuration was generated by
 # `rubocop --auto-gen-config`
-# on 2016-01-07 14:25:59 -0800 using RuboCop version 0.34.2.
+# on 2016-01-13 15:12:03 -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
@@ -24,7 +24,7 @@ Lint/UselessAssignment:
     - 'spec/unit/indexer_spec.rb'
     - 'spec/unit/solr_doc_builder_spec.rb'
 
-# Offense count: 22
+# Offense count: 21
 Metrics/AbcSize:
   Max: 82
 
@@ -37,12 +37,12 @@ Metrics/ClassLength:
 Metrics/CyclomaticComplexity:
   Max: 9
 
-# Offense count: 309
+# Offense count: 318
 # Configuration parameters: AllowURI, URISchemes.
 Metrics/LineLength:
   Max: 258
 
-# Offense count: 16
+# Offense count: 15
 # Configuration parameters: CountComments.
 Metrics/MethodLength:
   Max: 43
@@ -70,12 +70,6 @@ RSpec/InstanceVariable:
     - 'spec/unit/public_xml_fields_spec.rb'
     - 'spec/unit/solr_doc_builder_spec.rb'
 
-# Offense count: 6
-# Cop supports --auto-correct.
-Style/AlignArray:
-  Exclude:
-    - 'lib/gdor/indexer/mods_fields.rb'
-
 # Offense count: 7
 # Cop supports --auto-correct.
 # Configuration parameters: EnforcedStyle, SupportedStyles.
@@ -106,7 +100,7 @@ Style/DoubleNegation:
   Exclude:
     - 'lib/gdor/indexer/solr_doc_hash.rb'
 
-# Offense count: 2
+# Offense count: 1
 # Cop supports --auto-correct.
 # Configuration parameters: EnforcedStyle, SupportedStyles.
 Style/EmptyLinesAroundBlockBody:
@@ -119,7 +113,7 @@ Style/GuardClause:
   Exclude:
     - 'lib/gdor/indexer.rb'
 
-# Offense count: 14
+# Offense count: 21
 # Cop supports --auto-correct.
 # Configuration parameters: EnforcedStyle, SupportedStyles.
 Style/MultilineOperationIndentation:

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', '>= 1.3.4' # for new pub date methods
+  spec.add_dependency 'stanford-mods', '>= 1.4.0' # for new pub date methods
   spec.add_dependency 'nokogiri'
   spec.add_dependency 'rsolr'
   spec.add_dependency 'activesupport'

data/lib/gdor/indexer/mods_fields.rb

@@ -4,7 +4,7 @@ module GDor::Indexer::ModsFields
   # Create a Hash representing a Solr doc, with all MODS related fields populated.
   # @return [Hash] Hash representing the Solr document
   def doc_hash_from_mods
-    sort_str_w_approx_dates = smods_rec.pub_date_sortable_string(false)
+    pub_year_w_approx_dates = smods_rec.pub_year_int(false)
     doc_hash = {
       # title fields
       title_245a_search: smods_rec.sw_short_title,
@@ -46,22 +46,22 @@ module GDor::Indexer::ModsFields
 
       # publication fields
       pub_search: smods_rec.place,
-      pub_date_sort: sort_str_w_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_sortable_string(true),
-      pub_year_w_approx_isi: sort_str_w_approx_dates,
-      # TODO:  remove pub_date after reindexing existing colls;  deprecated in favor of pub_year_xxx_approx_isi ...
-      pub_date: smods_rec.pub_date_facet,
-      # display fields
+      pub_year_isi: pub_year_w_approx_dates, # for sorting
+      # TODO:  remove pub_date_sort after reindexing existing colls;  deprecated in favor of pub_year_isi ...
+      pub_date_sort: smods_rec.pub_year_sort_str(false),
+      # these are for single value facet display (in leiu of date slider (pub_year_tisim) )
+      pub_year_no_approx_isi: smods_rec.pub_year_int(true),
+      pub_year_w_approx_isi: pub_year_w_approx_dates,
+      # display fields  TODO:  pub_date_display is deprecated;  need better implementation of imprint_display
       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)),
+      creation_year_isi: smods_rec.year_int(smods_rec.date_created_elements(false)),
+      publication_year_isi: smods_rec.year_int(smods_rec.date_issued_elements(false)),
 
       all_search: smods_rec.text.gsub(/\s+/, ' ')
     }
 
-    more_pub_date_goodness(doc_hash)
+    add_pub_year_tisim(doc_hash)
   end
 
   private
@@ -76,30 +76,13 @@ module GDor::Indexer::ModsFields
     vals
   end
 
-  # additional pub date field processing for Solr doc hash
+  # add pub_year_tisim (for Date Slider; positive values only) to doc_hash
   # @param [Hash] Hash representing the Solr document
   # @return [Hash] updated Hash representing the Solr document
-  def more_pub_date_goodness(doc_hash)
-    pub_date_sort_val = doc_hash[:pub_date_sort]
+  def add_pub_year_tisim(doc_hash)
+    pub_date_sort_val = doc_hash[:pub_year_isi]
     if is_positive_int? pub_date_sort_val
       doc_hash[:pub_year_tisim] = pub_date_sort_val # for date slider
-      # remove leading zeros
-      [:pub_year_no_approx_isi,
-        :pub_year_w_approx_isi,
-        :creation_year_isi,
-        :publication_year_isi
-      ].each do |field_sym|
-        doc_hash[field_sym] = remove_leading_zeros(doc_hash[field_sym]) if doc_hash[field_sym]
-      end
-    else
-      # turn B.C. into -yyy for facet/display fields
-      [:pub_year_no_approx_isi,
-        :pub_year_w_approx_isi,
-        :creation_year_isi,
-        :publication_year_isi
-      ].each do |field_sym|
-        doc_hash[field_sym] = '-' + (1000 + doc_hash[field_sym].to_i).to_s if doc_hash[field_sym]
-      end
     end
     doc_hash
   end
@@ -110,10 +93,4 @@ module GDor::Indexer::ModsFields
   rescue
     false
   end
-
-  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 = '0.4.1'
+    VERSION = '0.5.0'
   end
 end

data/spec/unit/mods_pub_fields_spec.rb

@@ -23,7 +23,7 @@ describe GDor::Indexer::ModsFields do
 
   context 'publication date fields' do
 
-    RSpec.shared_examples 'expected' do |solr_field_sym, mods_field_val, exp_val|
+    RSpec.shared_examples 'expected (dateIssued)' do |solr_field_sym, mods_field_val, exp_val|
       it "#{exp_val} for #{mods_field_val}" do
         m = mods_origin_info_start_str +
               "<dateIssued>#{mods_field_val}</dateIssued>" +
@@ -33,51 +33,57 @@ describe GDor::Indexer::ModsFields do
       end
     end
 
-    context 'pub_date (to know current behavior)' do
-      it 'calls Stanford::Mods::Record instance pub_date_sortable_string(false)' do
-        expect(sdb.smods_rec).to receive(:pub_date_facet)
-        sdb.doc_hash_from_mods[:pub_date]
+    RSpec.shared_examples 'pub year int (dateIssued)' do |solr_field_sym|
+      it_behaves_like 'expected (dateIssued)', solr_field_sym, '1945', 1945
+      it_behaves_like 'expected (dateIssued)', solr_field_sym, '945', 945
+      it_behaves_like 'expected (dateIssued)', solr_field_sym, '66', 66
+      it_behaves_like 'expected (dateIssued)', solr_field_sym, '5', 5
+      it_behaves_like 'expected (dateIssued)', solr_field_sym, '0', 0
+      it_behaves_like 'expected (dateIssued)', solr_field_sym, 'blah blah 1945 blah blah', 1945
+      it_behaves_like 'expected (dateIssued)', solr_field_sym, '16--', 1600
+      it_behaves_like 'expected (dateIssued)', solr_field_sym, '9--', 900
+      it_behaves_like 'expected (dateIssued)', solr_field_sym, '19th century', 1800
+      it_behaves_like 'expected (dateIssued)', solr_field_sym, '9th century', 800
+      it_behaves_like 'expected (dateIssued)', solr_field_sym, 'blah June 4, 1594; blah 1596', 1594
+      it_behaves_like 'expected (dateIssued)', solr_field_sym, 'Aug. 3rd, 1886', 1886
+      it_behaves_like 'expected (dateIssued)', solr_field_sym, 'Aug. 3rd, [18]86? foo bar', 1886
+      it_behaves_like 'expected (dateIssued)', solr_field_sym, 'early 1890s', 1890
+      it_behaves_like 'expected (dateIssued)', solr_field_sym, '1865-6', 1865
+    end
+    RSpec.shared_examples 'pub year int (dateIssued) for B.C. years' do |solr_field_sym|
+      it_behaves_like 'expected (dateIssued)', solr_field_sym, '-4', -4
+      it_behaves_like 'expected (dateIssued)', solr_field_sym, '-15', -15
+      it_behaves_like 'expected (dateIssued)', solr_field_sym, '-666', -666
+      it_behaves_like 'expected (dateIssued)', solr_field_sym, '300 B.C.', -300
+    end
+
+    context 'pub_year_isi' do
+      it 'calls Stanford::Mods::Record instance pub_year_int(false)' do
+        expect(sdb.smods_rec).to receive(:pub_year_int).with(false)
+        allow(sdb.smods_rec).to receive(:pub_year_int).with(true) # pub_year_no_approx_isi
+        sdb.doc_hash_from_mods[:pub_year_isi]
       end
       it 'includes approx dates' do
         m = mods_origin_info_start_str +
               "<dateIssued qualifier='approximate'>1945</dateIssued>" +
             mods_origin_info_end_str
         sdb = sdb_for_mods(m)
-        expect(sdb.doc_hash_from_mods[:pub_date]).to eq('1945')
+        expect(sdb.doc_hash_from_mods[:pub_year_isi]).to eq 1945
       end
-      it 'takes single dateCreated' do
+      it 'will use dateCreated' do
         m = mods_origin_info_start_str +
               "<dateCreated>1904</dateCreated>" +
             mods_origin_info_end_str
         sdb = sdb_for_mods(m)
-        expect(sdb.doc_hash_from_mods[:pub_date]).to eq('1904')
+        expect(sdb.doc_hash_from_mods[:pub_year_isi]).to eq 1904
       end
-      it_behaves_like 'expected', :pub_date, 'blah blah 1945 blah', '1945'
-      it_behaves_like 'expected', :pub_date, '1945', '1945'
-      it_behaves_like 'expected', :pub_date, '945', '945'
-      it_behaves_like 'expected', :pub_date, '66', nil
-      it_behaves_like 'expected', :pub_date, '5', nil
-      it_behaves_like 'expected', :pub_date, '0', nil
-      it_behaves_like 'expected', :pub_date, '-4', nil
-      it_behaves_like 'expected', :pub_date, '-15', nil
-      it_behaves_like 'expected', :pub_date, '-666', '666' # WRONG
-      it_behaves_like 'expected', :pub_date, '16--', nil
-      it_behaves_like 'expected', :pub_date, '8--', nil
-      it_behaves_like 'expected', :pub_date, '19th century', '19th century'
-      it_behaves_like 'expected', :pub_date, '9th century', '9th century'
-      it_behaves_like 'expected', :pub_date, '300 B.C.', '300 B.C.'
-      it_behaves_like 'expected', :pub_date, 'Text dated June 4, 1594; miniatures added by 1596', '1594'
-      it_behaves_like 'expected', :pub_date, 'Aug. 3rd, 1886', '1886'
-      it_behaves_like 'expected', :pub_date, 'Aug. 3rd, [18]86?', '1886'
-      it_behaves_like 'expected', :pub_date, 'early 1890s', '1890'
-      it_behaves_like 'expected', :pub_date, '1865-6', '1865'
-
+      it_behaves_like 'pub year int (dateIssued)', :pub_year_isi
+      it_behaves_like 'pub year int (dateIssued) for B.C. years', :pub_year_isi
     end
 
-    context 'pub_date_sort' do
-      it 'calls Stanford::Mods::Record instance pub_date_sortable_string(false)' do
-        expect(sdb.smods_rec).to receive(:pub_date_sortable_string).with(false)
-        allow(sdb.smods_rec).to receive(:pub_date_sortable_string).with(true) # for pub_year_no_approx_isi
+    context 'pub_date_sort (DEPRECATED in favor of pub_year_isi)' do
+      it 'calls Stanford::Mods::Record instance pub_year_sort_str(false)' do
+        expect(sdb.smods_rec).to receive(:pub_year_sort_str).with(false)
         sdb.doc_hash_from_mods[:pub_date_sort]
       end
       it 'includes approx dates' do
@@ -94,74 +100,88 @@ describe GDor::Indexer::ModsFields do
         sdb = sdb_for_mods(m)
         expect(sdb.doc_hash_from_mods[:pub_date_sort]).to eq('1904')
       end
-      it_behaves_like 'expected', :pub_date_sort, '1945', '1945'
-      it_behaves_like 'expected', :pub_date_sort, '945', '0945'
-      it_behaves_like 'expected', :pub_date_sort, '66', '0066'
-      it_behaves_like 'expected', :pub_date_sort, '5', '0005'
-      it_behaves_like 'expected', :pub_date_sort, '0', '0000'
+      it_behaves_like 'expected (dateIssued)', :pub_date_sort, '1945', '1945'
+      it_behaves_like 'expected (dateIssued)', :pub_date_sort, '945', '0945'
+      it_behaves_like 'expected (dateIssued)', :pub_date_sort, '66', '0066'
+      it_behaves_like 'expected (dateIssued)', :pub_date_sort, '5', '0005'
+      it_behaves_like 'expected (dateIssued)', :pub_date_sort, '0', '0000'
+      it_behaves_like 'expected (dateIssued)', :pub_date_sort, '16--', '1600'
+      it_behaves_like 'expected (dateIssued)', :pub_date_sort, '9--', '0900'
+      it_behaves_like 'expected (dateIssued)', :pub_date_sort, '19th century', '1800'
+      it_behaves_like 'expected (dateIssued)', :pub_date_sort, '9th century', '0800'
       # these negative values are for String lexical sorting as this is a string
-      it_behaves_like 'expected', :pub_date_sort, '-4', '-996'
-      it_behaves_like 'expected', :pub_date_sort, '-15', '-985'
-      it_behaves_like 'expected', :pub_date_sort, '-666', '-334'
-      it_behaves_like 'expected', :pub_date_sort, '16--', '1600'
-      it_behaves_like 'expected', :pub_date_sort, '9--', '0900'
-      it_behaves_like 'expected', :pub_date_sort, '19th century', '1800'
-      it_behaves_like 'expected', :pub_date_sort, '9th century', '0800'
       # -(1000 - |yyy|) for BC dates
-      it_behaves_like 'expected', :pub_date_sort, '300 B.C.', '-700'
+      it_behaves_like 'expected (dateIssued)', :pub_date_sort, '-4', '-996'
+      it_behaves_like 'expected (dateIssued)', :pub_date_sort, '-15', '-985'
+      it_behaves_like 'expected (dateIssued)', :pub_date_sort, '-666', '-334'
+      it_behaves_like 'expected (dateIssued)', :pub_date_sort, '300 B.C.', '-700'
     end
 
-    context 'single valued pub year facets' do
-      let(:mods) do
-        mods_origin_info_start_str +
-          "<dateIssued qualifier=\"approximate\">1500</dateIssued>
-          <dateIssued>2000</dateIssued>" +
-        mods_origin_info_end_str
+    context 'pub_year_no_approx_isi' do
+      it 'does not include approx dates' do
+        m = mods_origin_info_start_str +
+              "<dateIssued qualifier='approximate'>1945</dateIssued>" +
+          mods_origin_info_end_str
+        sdb = sdb_for_mods(m)
+        expect(sdb.doc_hash_from_mods[:pub_year_no_approx_isi]).to eq nil
       end
-      it 'pub_year_no_approx_isi calls Stanford::Mods::Record instance pub_date_facet_single_value(true)' do
+      it 'calls Stanford::Mods::Record instance pub_year_int(true)' do
+        mods = mods_origin_info_start_str +
+                "<dateIssued qualifier=\"approximate\">1500</dateIssued>
+                <dateIssued>2000</dateIssued>" +
+              mods_origin_info_end_str
         sdb = sdb_for_mods(mods)
-        expect(sdb.smods_rec).to receive(:pub_date_sortable_string).with(true).and_call_original
-        allow(sdb.smods_rec).to receive(:pub_date_sortable_string).with(false) # for other flavor
-        expect(sdb.doc_hash_from_mods[:pub_year_no_approx_isi]).to eq '2000'
+        expect(sdb.smods_rec).to receive(:pub_year_int).with(true).and_call_original
+        allow(sdb.smods_rec).to receive(:pub_year_int).with(false) # for other flavor
+        expect(sdb.doc_hash_from_mods[:pub_year_no_approx_isi]).to eq 2000
+      end
+      it 'will use dateCreated' do
+        m = mods_origin_info_start_str +
+              "<dateCreated>1904</dateCreated>" +
+            mods_origin_info_end_str
+        sdb = sdb_for_mods(m)
+        expect(sdb.doc_hash_from_mods[:pub_year_no_approx_isi]).to eq 1904
       end
-      it 'pub_year_w_approx_isi calls Stanford::Mods::Record instance pub_date_facet_single_value(false)' do
+      it_behaves_like 'pub year int (dateIssued)', :pub_year_no_approx_isi
+      it_behaves_like 'pub year int (dateIssued) for B.C. years', :pub_year_no_approx_isi
+    end
+
+    context 'pub_year_w_approx_isi' do
+      it 'includes approx dates' do
+        m = mods_origin_info_start_str +
+              "<dateIssued qualifier='approximate'>1945</dateIssued>" +
+          mods_origin_info_end_str
+        sdb = sdb_for_mods(m)
+        expect(sdb.doc_hash_from_mods[:pub_year_w_approx_isi]).to eq 1945
+      end
+      it 'calls Stanford::Mods::Record instance pub_year_int(false)' do
+        mods = mods_origin_info_start_str +
+                "<dateIssued qualifier=\"approximate\">1500</dateIssued>
+                <dateIssued>2000</dateIssued>" +
+              mods_origin_info_end_str
         sdb = sdb_for_mods(mods)
-        expect(sdb.smods_rec).to receive(:pub_date_sortable_string).with(false).and_call_original
-        allow(sdb.smods_rec).to receive(:pub_date_sortable_string).with(true) # for other flavor
-        expect(sdb.doc_hash_from_mods[:pub_year_w_approx_isi]).to eq '1500'
+        expect(sdb.smods_rec).to receive(:pub_year_int).with(false).and_call_original
+        allow(sdb.smods_rec).to receive(:pub_year_int).with(true) # for other flavor
+        expect(sdb.doc_hash_from_mods[:pub_year_w_approx_isi]).to eq 1500
       end
-      RSpec.shared_examples "single pub year facet" do |field_sym|
-        it_behaves_like 'expected', field_sym, 'blah blah 1945 blah', '1945'
-        it_behaves_like 'expected', field_sym, '1945', '1945'
-        it_behaves_like 'expected', field_sym, '945', '945'
-        it_behaves_like 'expected', field_sym, '66', '66'
-        it_behaves_like 'expected', field_sym, '5', '5'
-        it_behaves_like 'expected', field_sym, '0', '0'
-        it_behaves_like 'expected', field_sym, '-4', '-4'
-        it_behaves_like 'expected', field_sym, '-15', '-15'
-        it_behaves_like 'expected', field_sym, '-666', '-666'
-        it_behaves_like 'expected', field_sym, '16--', '1600'
-        it_behaves_like 'expected', field_sym, '8--', '800'
-        it_behaves_like 'expected', field_sym, '19th century', '1800'
-        it_behaves_like 'expected', field_sym, '9th century', '800'
-        it_behaves_like 'expected', field_sym, '300 B.C.', '-300'
-        it_behaves_like 'expected', field_sym, 'Text dated June 4, 1594; miniatures added by 1596', '1594'
-        it_behaves_like 'expected', field_sym, 'Aug. 3rd, 1886', '1886'
-        it_behaves_like 'expected', field_sym, 'Aug. 3rd, [18]86?', '1886'
-        it_behaves_like 'expected', field_sym, 'early 1890s', '1890'
-        it_behaves_like 'expected', field_sym, '1865-6', '1865'
+      it 'will use dateCreated' do
+        m = mods_origin_info_start_str +
+              "<dateCreated>1904</dateCreated>" +
+            mods_origin_info_end_str
+        sdb = sdb_for_mods(m)
+        expect(sdb.doc_hash_from_mods[:pub_year_w_approx_isi]).to eq 1904
       end
-      it_behaves_like "single pub year facet", :pub_year_no_approx_isi
-      it_behaves_like "single pub year facet", :pub_year_w_approx_isi
+      it_behaves_like 'pub year int (dateIssued)', :pub_year_w_approx_isi
+      it_behaves_like 'pub year int (dateIssued) for B.C. years', :pub_year_w_approx_isi
     end
 
     context 'pub_year_tisim for date slider' do
       it 'should handle multiple pub dates (to be implemented - esp for date slider)'
 
       # FIXME:  it should be using a method approp for date slider values, not single value
-      it 'pub_year_tisim calls Stanford::Mods::Record instance pub_date_sortable_string(false)' do
-        expect(sdb.smods_rec).to receive(:pub_date_sortable_string).with(false)
-        allow(sdb.smods_rec).to receive(:pub_date_sortable_string).with(true) # for pub_year_no_approx_isi
+      it 'pub_year_tisim calls Stanford::Mods::Record instance pub_year_int(false)' do
+        expect(sdb.smods_rec).to receive(:pub_year_int).with(false)
+        allow(sdb.smods_rec).to receive(:pub_year_int).with(true) # for pub_year_no_approx_isi
         sdb.doc_hash_from_mods[:pub_year_tisim]
       end
       it 'includes approx dates' do
@@ -169,53 +189,36 @@ describe GDor::Indexer::ModsFields do
               "<dateIssued qualifier='approximate'>1945</dateIssued>" +
           mods_origin_info_end_str
         sdb = sdb_for_mods(m)
-        expect(sdb.doc_hash_from_mods[:pub_year_tisim]).to eq('1945')
+        expect(sdb.doc_hash_from_mods[:pub_year_tisim]).to eq 1945
       end
-      it 'takes single dateCreated' do
+      it 'will use dateCreated' do
         m = mods_origin_info_start_str +
               "<dateCreated>1904</dateCreated>" +
             mods_origin_info_end_str
         sdb = sdb_for_mods(m)
-        expect(sdb.doc_hash_from_mods[:pub_year_tisim]).to eq('1904')
-      end
-      it 'ignores B.C. dates' do
-        m = mods_origin_info_start_str +
-              "<dateCreated>300 B.C.</dateCreated>" +
-          mods_origin_info_end_str
-        sdb = sdb_for_mods(m)
-        expect(sdb.doc_hash_from_mods).not_to have_key(:pub_year_tisim)
-        expect(sdb.doc_hash_from_mods[:pub_year_tisim]).to be_nil
+        expect(sdb.doc_hash_from_mods[:pub_year_tisim]).to eq 1904
       end
-      it_behaves_like 'expected', :pub_year_tisim, '1945', '1945'
-      it_behaves_like 'expected', :pub_year_tisim, '945', '0945'
-      it_behaves_like 'expected', :pub_year_tisim, '66', '0066'
-      it_behaves_like 'expected', :pub_year_tisim, '5', '0005'
-      it_behaves_like 'expected', :pub_year_tisim, '0', '0000'
-      it_behaves_like 'expected', :pub_year_tisim, '-4', nil
-      it_behaves_like 'expected', :pub_year_tisim, '-15', nil
-      it_behaves_like 'expected', :pub_year_tisim, '-666', nil
-      it_behaves_like 'expected', :pub_year_tisim, '16--', '1600'
-      it_behaves_like 'expected', :pub_year_tisim, '9--', '0900'
-      it_behaves_like 'expected', :pub_year_tisim, '19th century', '1800'
-      it_behaves_like 'expected', :pub_year_tisim, '9th century', '0800'
-      it_behaves_like 'expected', :pub_year_tisim, 'Text dated June 4, 1594; miniatures added by 1596', '1594'
-      it_behaves_like 'expected', :pub_year_tisim, 'Aug. 3rd, 1886', '1886'
-      it_behaves_like 'expected', :pub_year_tisim, 'Aug. 3rd, [18]86?', '1886'
-      it_behaves_like 'expected', :pub_year_tisim, 'early 1890s', '1890'
-      it_behaves_like 'expected', :pub_year_tisim, '1865-6', '1865' # FIXME:  should be both years
+      it_behaves_like 'pub year int (dateIssued)', :pub_year_tisim
+      # nil for years before 0
+      it_behaves_like 'expected (dateIssued)', :pub_year_tisim, '-4', nil
+      it_behaves_like 'expected (dateIssued)', :pub_year_tisim, '-15', nil
+      it_behaves_like 'expected (dateIssued)', :pub_year_tisim, '-666', nil
+      it_behaves_like 'expected (dateIssued)', :pub_year_tisim, '300 B.C.', nil
+      it_behaves_like 'expected (dateIssued)', :pub_year_tisim, 'blah June 4, 1594; blah 1596', 1594 # FIXME:  should be both years
+      it_behaves_like 'expected (dateIssued)', :pub_year_tisim, '1865-6', 1865 # FIXME:  should be both years
     end
 
     context 'creation_year_isi' do
-      it 'creation_year_isi calls Stanford::Mods::Record pub_date_best_sort_str_value for dateCreated elements' do
+      it 'creation_year_isi calls Stanford::Mods::Record year_int for dateCreated elements' do
         m = mods_origin_info_start_str +
               "<dateCreated qualifier='approximate'>1500</dateCreated>
               <dateIssued qualifier='approximate'>2000</dateIssued>" +
             mods_origin_info_end_str
         sdb = sdb_for_mods(m)
-        expect(sdb.smods_rec).to receive(:pub_date_best_sort_str_value).at_least(2).times.and_call_original
-        expect(sdb.doc_hash_from_mods[:creation_year_isi]).to eq '1500'
+        expect(sdb.smods_rec).to receive(:year_int).at_least(2).times.and_call_original
+        expect(sdb.doc_hash_from_mods[:creation_year_isi]).to eq 1500
       end
-      RSpec.shared_examples 'expected for dateCreated' do |mods_field_val, exp_val|
+      RSpec.shared_examples 'expected (dateCreated)' do |mods_field_val, exp_val|
         it "#{exp_val} for #{mods_field_val}" do
           m = mods_origin_info_start_str +
                 "<dateCreated>#{mods_field_val}</dateCreated>" +
@@ -224,58 +227,39 @@ describe GDor::Indexer::ModsFields do
           expect(sdb.doc_hash_from_mods[:creation_year_isi]).to eq exp_val
         end
       end
-      it_behaves_like 'expected for dateCreated', '1945', '1945'
-      # note that it removes leading zeros
-      it_behaves_like 'expected for dateCreated', '945', '945'
-      it_behaves_like 'expected for dateCreated', '66', '66'
-      it_behaves_like 'expected for dateCreated', '5', '5'
-      it_behaves_like 'expected for dateCreated', '0', '0'
-      it_behaves_like 'expected for dateCreated', '-4', '-4'
-      it_behaves_like 'expected for dateCreated', '-15', '-15'
-      it_behaves_like 'expected for dateCreated', '-666', '-666'
-      it_behaves_like 'expected for dateCreated', '16--', '1600'
-      it_behaves_like 'expected for dateCreated', '9--', '900'
-      it_behaves_like 'expected for dateCreated', '19th century', '1800'
-      it_behaves_like 'expected for dateCreated', '9th century', '800'
-      it_behaves_like 'expected for dateCreated', 'blah June 4, 1594; blah 1596', '1594'
-      it_behaves_like 'expected for dateCreated', 'Aug. 3rd, 1886', '1886'
-      it_behaves_like 'expected for dateCreated', 'Aug. 3rd, [18]86?', '1886'
-      it_behaves_like 'expected for dateCreated', 'early 1890s', '1890'
-      it_behaves_like 'expected for dateCreated', '1865-6', '1865'
-      # note: B.C. becomes a regular negative number
-      it_behaves_like 'expected for dateCreated', '300 B.C.', '-300'
+      it_behaves_like 'expected (dateCreated)', '1945', 1945
+      it_behaves_like 'expected (dateCreated)', '945', 945
+      it_behaves_like 'expected (dateCreated)', '66', 66
+      it_behaves_like 'expected (dateCreated)', '5', 5
+      it_behaves_like 'expected (dateCreated)', '0', 0
+      it_behaves_like 'expected (dateCreated)', '16--', 1600
+      it_behaves_like 'expected (dateCreated)', '9--', 900
+      it_behaves_like 'expected (dateCreated)', 'blah blah 1945 blah blah', 1945
+      it_behaves_like 'expected (dateCreated)', '19th century', 1800
+      it_behaves_like 'expected (dateCreated)', '9th century', 800
+      it_behaves_like 'expected (dateCreated)', 'blah June 4, 1594; blah 1596', 1594
+      it_behaves_like 'expected (dateCreated)', 'Aug. 3rd, 1886', 1886
+      it_behaves_like 'expected (dateCreated)', 'Aug. 3rd, [18]86? foo bar', 1886
+      it_behaves_like 'expected (dateCreated)', 'early 1890s', 1890
+      it_behaves_like 'expected (dateCreated)', '1865-6', 1865
+      it_behaves_like 'expected (dateCreated)', '-4', -4
+      it_behaves_like 'expected (dateCreated)', '-15', -15
+      it_behaves_like 'expected (dateCreated)', '-666', -666
+      it_behaves_like 'expected (dateCreated)', '300 B.C.', -300
     end
 
     context 'publication_year_isi' do
-      it 'publication_year_isi calls Stanford::Mods::Record pub_date_best_sort_str_value for dateIssued elements' do
+      it 'publication_year_isi calls Stanford::Mods::Record year_int for dateIssued elements' do
         m = mods_origin_info_start_str +
               "<dateCreated qualifier='approximate'>1500</dateCreated>
               <dateIssued qualifier='approximate'>2000</dateIssued>" +
             mods_origin_info_end_str
         sdb = sdb_for_mods(m)
-        expect(sdb.smods_rec).to receive(:pub_date_best_sort_str_value).at_least(2).times.and_call_original
-        expect(sdb.doc_hash_from_mods[:publication_year_isi]).to eq '2000'
+        expect(sdb.smods_rec).to receive(:year_int).at_least(2).times.and_call_original
+        expect(sdb.doc_hash_from_mods[:publication_year_isi]).to eq 2000
       end
-      it_behaves_like 'expected', :publication_year_isi, '1945', '1945'
-      # note that it removes leading zeros
-      it_behaves_like 'expected', :publication_year_isi, '945', '945'
-      it_behaves_like 'expected', :publication_year_isi, '66', '66'
-      it_behaves_like 'expected', :publication_year_isi, '5', '5'
-      it_behaves_like 'expected', :publication_year_isi, '0', '0'
-      it_behaves_like 'expected', :publication_year_isi, '-4', '-4'
-      it_behaves_like 'expected', :publication_year_isi, '-15', '-15'
-      it_behaves_like 'expected', :publication_year_isi, '-666', '-666'
-      it_behaves_like 'expected', :publication_year_isi, '16--', '1600'
-      it_behaves_like 'expected', :publication_year_isi, '9--', '900'
-      it_behaves_like 'expected', :publication_year_isi, '19th century', '1800'
-      it_behaves_like 'expected', :publication_year_isi, '9th century', '800'
-      it_behaves_like 'expected', :publication_year_isi, 'blah June 4, 1594; blah 1596', '1594'
-      it_behaves_like 'expected', :publication_year_isi, 'Aug. 3rd, 1886', '1886'
-      it_behaves_like 'expected', :publication_year_isi, 'Aug. 3rd, [18]86?', '1886'
-      it_behaves_like 'expected', :publication_year_isi, 'early 1890s', '1890'
-      it_behaves_like 'expected', :publication_year_isi, '1865-6', '1865'
-      # note: B.C. becomes a regular negative number
-      it_behaves_like 'expected', :publication_year_isi, '300 B.C.', '-300'
+      it_behaves_like 'pub year int (dateIssued)', :publication_year_isi
+      it_behaves_like 'pub year int (dateIssued) for B.C. years', :publication_year_isi
     end
   end # publication date fields
 
@@ -285,22 +269,22 @@ describe GDor::Indexer::ModsFields do
       expect(sdb.smods_rec).to receive(:pub_date_display)
       sdb.doc_hash_from_mods[:imprint_display]
     end
-    it_behaves_like 'expected', :imprint_display, '1945', '1945'
-    it_behaves_like 'expected', :imprint_display, '945', '945'
-    it_behaves_like 'expected', :imprint_display, '66', '66'
-    it_behaves_like 'expected', :imprint_display, '5', '5'
-    it_behaves_like 'expected', :imprint_display, '0', '0'
-    it_behaves_like 'expected', :imprint_display, '-4', '-4'
-    it_behaves_like 'expected', :imprint_display, '-15', '-15'
-    it_behaves_like 'expected', :imprint_display, '-666', '-666'
-    it_behaves_like 'expected', :imprint_display, '16--', '16--'
-    it_behaves_like 'expected', :imprint_display, '9--', '9--'
-    it_behaves_like 'expected', :imprint_display, '19th century', '19th century'
-    it_behaves_like 'expected', :imprint_display, '9th century', '9th century'
-    it_behaves_like 'expected', :imprint_display, 'blah June 4, 1594; blah 1596', 'blah June 4, 1594; blah 1596'
-    it_behaves_like 'expected', :imprint_display, 'Aug. 3rd, 1886', 'Aug. 3rd, 1886'
-    it_behaves_like 'expected', :imprint_display, 'Aug. 3rd, [18]86?', 'Aug. 3rd, [18]86?'
-    it_behaves_like 'expected', :imprint_display, 'early 1890s', 'early 1890s'
-    it_behaves_like 'expected', :imprint_display, '1865-6', '1865-6'
+    it_behaves_like 'expected (dateIssued)', :imprint_display, '1945', '1945'
+    it_behaves_like 'expected (dateIssued)', :imprint_display, '945', '945'
+    it_behaves_like 'expected (dateIssued)', :imprint_display, '66', '66'
+    it_behaves_like 'expected (dateIssued)', :imprint_display, '5', '5'
+    it_behaves_like 'expected (dateIssued)', :imprint_display, '0', '0'
+    it_behaves_like 'expected (dateIssued)', :imprint_display, '-4', '-4'
+    it_behaves_like 'expected (dateIssued)', :imprint_display, '-15', '-15'
+    it_behaves_like 'expected (dateIssued)', :imprint_display, '-666', '-666'
+    it_behaves_like 'expected (dateIssued)', :imprint_display, '16--', '16--'
+    it_behaves_like 'expected (dateIssued)', :imprint_display, '9--', '9--'
+    it_behaves_like 'expected (dateIssued)', :imprint_display, '19th century', '19th century'
+    it_behaves_like 'expected (dateIssued)', :imprint_display, '9th century', '9th century'
+    it_behaves_like 'expected (dateIssued)', :imprint_display, 'blah June 4, 1594; blah 1596', 'blah June 4, 1594; blah 1596'
+    it_behaves_like 'expected (dateIssued)', :imprint_display, 'Aug. 3rd, 1886', 'Aug. 3rd, 1886'
+    it_behaves_like 'expected (dateIssued)', :imprint_display, 'Aug. 3rd, [18]86?', 'Aug. 3rd, [18]86?'
+    it_behaves_like 'expected (dateIssued)', :imprint_display, 'early 1890s', 'early 1890s'
+    it_behaves_like 'expected (dateIssued)', :imprint_display, '1865-6', '1865-6'
   end
 end

metadata

@@ -1,7 +1,7 @@
 --- !ruby/object:Gem::Specification
 name: gdor-indexer
 version: !ruby/object:Gem::Version
-  version: 0.4.1
+  version: 0.5.0
 platform: ruby
 authors:
 - Naomi Dushay
@@ -10,7 +10,7 @@ authors:
 autorequire: 
 bindir: bin
 cert_chain: []
-date: 2016-01-07 00:00:00.000000000 Z
+date: 2016-01-19 00:00:00.000000000 Z
 dependencies:
 - !ruby/object:Gem::Dependency
   name: harvestdor-indexer
@@ -32,14 +32,14 @@ dependencies:
     requirements:
     - - ">="
       - !ruby/object:Gem::Version
-        version: 1.3.4
+        version: 1.4.0
   type: :runtime
   prerelease: false
   version_requirements: !ruby/object:Gem::Requirement
     requirements:
     - - ">="
       - !ruby/object:Gem::Version
-        version: 1.3.4
+        version: 1.4.0
 - !ruby/object:Gem::Dependency
   name: nokogiri
   requirement: !ruby/object:Gem::Requirement