stanford-mods 2.6.4 → 3.1.0

data/spec/imprint_spec.rb

@@ -1,5 +1,5 @@
 def stanford_mods_imprint(smods_rec)
-  Stanford::Mods::Imprint.new(smods_rec.origin_info)
+  Stanford::Mods::Imprint.new(smods_rec.origin_info.first)
 end
 
 # unit / functional tests for imprint class
@@ -9,216 +9,272 @@ describe Stanford::Mods::Imprint do
   let(:mods_origin_info_end_str) { '</originInfo></mods>' }
 
   describe 'date processing' do
-    describe '#publication_date_for_slider' do
-      {
-        '' => [],
-        '<dateIssued point="start">unparsable</dateIssued>' => [],
-        '<dateIssued>1957</dateIssued>' => [1957],
-        '<dateCreated>1957</dateCreated>' => [1957],
-        '<dateIssued>195u</dateIssued>' => (1950..1959).to_a,
-        '<dateCreated keyDate="yes">1964</dateCreated><dateIssued>195u</dateIssued>' => [1964],
-        '<dateIssued>1964</dateIssued><dateCreated>195u</dateCreated>' => [1964],
-        '<dateIssued point="start">195u</dateIssued><dateIssued point="end">1964</dateIssued>' => (1950..1964).to_a,
-        '<dateIssued>1964</dateIssued><dateIssued>195u</dateIssued>' => [1964] + (1950..1959).to_a,
-        '<dateCreated keyDate="yes" encoding="w3cdtf" point="start">1966</dateCreated><dateCreated encoding="w3cdtf" point="end">1967</dateCreated>' => [1966,1967]
-      }.each do |example, expected|
-        it 'works' do
-          smods_rec.from_str("#{mods_origin_info_start_str}
-            #{example}
-          #{mods_origin_info_end_str}")
-          imprint = stanford_mods_imprint(smods_rec)
-          expect(imprint.publication_date_for_slider).to eq expected
-        end
-      end
+    it 'ignores bad dates' do
+      smods_rec.from_str(mods_origin_info_start_str +
+        '<place>
+            <placeTerm>Spain</placeTerm>
+          </place>
+          <dateIssued>
+            9999
+          </dateIssued>
+          <dateIssued>
+            uuuu
+          </dateIssued>
+          <dateIssued>
+            0000-00-00
+          </dateIssued>' +
+        mods_origin_info_end_str)
+      imp = stanford_mods_imprint(smods_rec)
+      updated_element = imp.send(:date_str)
+      expect(updated_element.to_s).to be_empty
     end
 
-    describe 'bad dates' do
-      it 'ignores bad date values' do
-        smods_rec.from_str(mods_origin_info_start_str +
-          '<place>
-             <placeTerm>Spain</placeTerm>
-           </place>
-           <dateIssued>
-             9999
-           </dateIssued>
-           <dateIssued>
-             uuuu
-           </dateIssued>
-           <dateIssued>
-             0000-00-00
-           </dateIssued>' +
-          mods_origin_info_end_str)
-        imprint_strs = stanford_mods_imprint(smods_rec).imprint_statements
-        expect(imprint_strs).to eq(['Spain'])
-      end
-      it 'handles invalid dates by returning the original value' do
-        smods_rec.from_str(mods_origin_info_start_str +
-          '<dateCreated encoding="w3cdtf">1920-09-00</dateCreated>' +
-          mods_origin_info_end_str)
-        imprint_strs = stanford_mods_imprint(smods_rec).imprint_statements
-        expect(imprint_strs.first).to match('1920-09-00')
-      end
+    it 'handles invalid dates by returning the original value' do
+      smods_rec.from_str(mods_origin_info_start_str +
+        '<dateCreated encoding="w3cdtf">1920-09-32</dateCreated>' +
+        mods_origin_info_end_str)
+      imprint_strs = stanford_mods_imprint(smods_rec).display_str
+      expect(imprint_strs).to eq '1920-09-32'
     end
 
-    context '#process_decade_century_dates' do
-      it 'calls process_decade_date for each element in passed nodeset that matches decade str' do
-        smods_rec.from_str("#{mods_origin_info_start_str}
-          <dateIssued>before 195x after</dateIssued>
-          <dateIssued>1950s not a match</dateIssued>
-          <dateIssued>before 17-- after</dateIssued>
-          <dateIssued>not a match</dateIssued>
-          <dateIssued>another 165x match</dateIssued>
-        #{mods_origin_info_end_str}")
-        imp = stanford_mods_imprint(smods_rec)
-        dt_elements = smods_rec.origin_info.dateIssued
-        expect(imp).to receive(:date_is_decade?).and_return(true, false, false, false, true)
-        expect(imp).to receive(:process_decade_date).twice
-        expect(imp).to receive(:date_is_century?).and_return(false, true, false)
-        expect(imp).to receive(:process_century_date)
-        imp.send(:process_decade_century_dates, dt_elements)
-      end
+    it 'collects all the dates from date elements in the MODS' do
+      smods_rec.from_str <<-XML
+        #{mods_origin_info_start_str}
+          <dateIssued>17uu</dateIssued>
+          <dateIssued>ain't no date heah</dateIssued>
+          <dateCreated encoding="w3cdtf">1920-09</dateCreated>
+        #{mods_origin_info_end_str}
+      XML
+
+      imp = stanford_mods_imprint(smods_rec)
+      updated_element = imp.send(:date_str)
+      expect(updated_element).to include('18th century', "ain't no date heah", '1920')
     end
 
-    context 'date_is_decade?' do
-      ['156u',
-        '167-?]',
-        '[171-?]',
-        '[189-]',
-        'ca.170-?]',
-        '200-?]',
-        '186?',
-        '195x',
-        'before 195x after'
-      ].each do |example|
-        it 'true when decade string to change' do
-          smods_rec.from_str("#{mods_origin_info_start_str}
-            <dateIssued>#{example}</dateIssued>
-          #{mods_origin_info_end_str}")
-          imp = stanford_mods_imprint(smods_rec)
-          element = smods_rec.origin_info.dateIssued.first
-          expect(imp.send(:date_is_decade?, element)).to be_truthy
-        end
-      end
-      ['1950s',
-        "1950's",
-        'before 1950s after'
-      ].each do |example|
-        it 'false when no decade string to change' do
-          smods_rec.from_str("#{mods_origin_info_start_str}
-            <dateIssued>#{example}</dateIssued>
-          #{mods_origin_info_end_str}")
-          imp = stanford_mods_imprint(smods_rec)
-          element = smods_rec.origin_info.dateIssued.first
-          expect(imp.send(:date_is_decade?, element)).to be_falsey
-        end
-      end
+    it 'prefers unencoded dates in order to preserve punctuation-as-metadata' do
+      smods_rec.from_str <<-XML
+        #{mods_origin_info_start_str}
+          <dateIssued>1874]</dateIssued>
+          <dateIssued encoding="w3cdtf">1874</dateIssued>
+        #{mods_origin_info_end_str}
+      XML
+
+      imp = stanford_mods_imprint(smods_rec)
+      updated_element = imp.send(:date_str)
+      expect(updated_element).to eq '1874]'
     end
 
-    context '#process_decade_date' do
-      {
-        '1950s' => '1950s',
-        "1950's" => "1950's",
-        '156u' => '1560s',
-        '167-?]' => '1670s?]',
-        '[171-?]' => '[1710s?]',
-        '[189-]' => '[1890s]',
-        'ca.170-?]' => 'ca.1700s?]',
-        '200-?]' => '2000s?]',
-        '186?' => '1860s',
-        '195x' => '1950s',
-        'early 1890s' => 'early 1890s',
-        'before 195x after' => 'before 1950s after'
-      }.each do |example, expected|
-        it "#{expected} for #{example}" do
-          smods_rec.from_str("#{mods_origin_info_start_str}
-            <dateIssued>#{example}</dateIssued>
-          #{mods_origin_info_end_str}")
-          imp = stanford_mods_imprint(smods_rec)
-          updated_element = imp.send(:process_decade_date, smods_rec.origin_info.dateIssued.first)
-          expect(updated_element.text).to eq expected
-        end
-      end
-      it 'leaves text alone when date str but no decade' do
-        smods_rec.from_str("#{mods_origin_info_start_str}
-          <dateIssued>I think July 15, 1965 was a great day</dateIssued>
-        #{mods_origin_info_end_str}")
-        imp = stanford_mods_imprint(smods_rec)
-        updated_element = imp.send(:process_decade_date, smods_rec.origin_info.dateIssued.first)
-        expect(updated_element.text).to eq 'I think July 15, 1965 was a great day'
-      end
-      it 'leaves text alone when no date str' do
-        smods_rec.from_str("#{mods_origin_info_start_str}
-          <dateIssued>ain't no date heah</dateIssued>
-        #{mods_origin_info_end_str}")
-        imp = stanford_mods_imprint(smods_rec)
-        updated_element = imp.send(:process_decade_date, smods_rec.origin_info.dateIssued.first)
-        expect(updated_element.text).to eq "ain't no date heah"
-      end
+    it 'deduplicates dates so we do not show the same information repeatedly' do
+      smods_rec.from_str <<-XML
+        #{mods_origin_info_start_str}
+          <dateIssued>1874</dateIssued>
+          <dateIssued encoding="edtf">1874</dateIssued>
+          <dateIssued encoding="w3cdtf">1874</dateIssued>
+        #{mods_origin_info_end_str}
+      XML
+
+      imp = stanford_mods_imprint(smods_rec)
+      updated_element = imp.send(:date_str)
+      expect(updated_element).to eq '1874'
     end
 
-    context 'date_is_century?' do
-      ['17uu',
-        '17--?',
-        'before [16--] after'
-      ].each do |example|
-        it 'true when century string to change' do
-          smods_rec.from_str("#{mods_origin_info_start_str}
-            <dateIssued>#{example}</dateIssued>
-          #{mods_origin_info_end_str}")
-          imp = stanford_mods_imprint(smods_rec)
-          element = smods_rec.origin_info.dateIssued.first
-          expect(imp.send(:date_is_century?, element)).to be_truthy
-        end
-      end
-      ['18th century CE',
-        "before 5th century after"
-      ].each do |example|
-        it 'false when no century string to change' do
-          smods_rec.from_str("#{mods_origin_info_start_str}
-            <dateIssued>#{example}</dateIssued>
-          #{mods_origin_info_end_str}")
-          imp = stanford_mods_imprint(smods_rec)
-          element = smods_rec.origin_info.dateIssued.first
-          expect(imp.send(:date_is_century?, element)).to be_falsey
-        end
-      end
+    it 'also deduplicates date ranges against a single date' do
+      smods_rec.from_str <<-XML
+        #{mods_origin_info_start_str}
+          <dateIssued>1874]</dateIssued>
+          <dateIssued point="start" encoding="edtf">1874</dateIssued>
+          <dateIssued point="end" encoding="w3cdtf">1876</dateIssued>
+        #{mods_origin_info_end_str}
+      XML
+
+      imp = stanford_mods_imprint(smods_rec)
+      updated_element = imp.send(:date_str)
+      expect(updated_element).to eq '1874]'
     end
 
-    context '#process_century_date' do
-      {
-        '18th century CE' => '18th century CE',
-        '17uu' => '18th century',
-        '17--?]' => '18th century?]',
-        '17--]' => '18th century]',
-        '[17--]' => '[18th century]',
-        '[17--?]' => '[18th century?]',
-        'before 16uu after' => 'before 17th century after'
-      }.each do |example, expected|
-        it "#{expected} for #{example}" do
-          smods_rec.from_str("#{mods_origin_info_start_str}
-            <dateIssued>#{example}</dateIssued>
-          #{mods_origin_info_end_str}")
-          imp = stanford_mods_imprint(smods_rec)
-          updated_element = imp.send(:process_century_date, smods_rec.origin_info.dateIssued.first)
-          expect(updated_element.text).to eq expected
-        end
-      end
-      it 'leaves text alone when date str but no century' do
-        smods_rec.from_str("#{mods_origin_info_start_str}
-          <dateIssued>I think July 15, 1965 was a great day</dateIssued>
-        #{mods_origin_info_end_str}")
-        imp = stanford_mods_imprint(smods_rec)
-        updated_element = imp.send(:process_century_date, smods_rec.origin_info.dateIssued.first)
-        expect(updated_element.text).to eq 'I think July 15, 1965 was a great day'
-      end
-      it 'leaves text alone when no date str' do
-        smods_rec.from_str("#{mods_origin_info_start_str}
-          <dateIssued>ain't no date heah</dateIssued>
-        #{mods_origin_info_end_str}")
-        imp = stanford_mods_imprint(smods_rec)
-        updated_element = imp.send(:process_century_date, smods_rec.origin_info.dateIssued.first)
-        expect(updated_element.text).to eq "ain't no date heah"
-      end
+    it 'presents date ranges' do
+      smods_rec.from_str <<-XML
+        #{mods_origin_info_start_str}
+          <dateIssued point="start" encoding="edtf">1874</dateIssued>
+          <dateIssued point="end" encoding="w3cdtf">1876</dateIssued>
+        #{mods_origin_info_end_str}
+      XML
+
+      imp = stanford_mods_imprint(smods_rec)
+      updated_element = imp.send(:date_str)
+      expect(updated_element).to eq '1874 - 1876'
+    end
+
+    it 'presents centuries' do
+      smods_rec.from_str <<-XML
+        #{mods_origin_info_start_str}
+          <dateIssued encoding="edtf">18XX</dateIssued>
+        #{mods_origin_info_end_str}
+      XML
+
+      imp = stanford_mods_imprint(smods_rec)
+      updated_element = imp.send(:date_str)
+      expect(updated_element).to eq '19th century'
+    end
+
+    it 'presents decades' do
+      smods_rec.from_str <<-XML
+        #{mods_origin_info_start_str}
+          <dateIssued encoding="edtf">147X</dateIssued>
+        #{mods_origin_info_end_str}
+      XML
+
+      imp = stanford_mods_imprint(smods_rec)
+      updated_element = imp.send(:date_str)
+      expect(updated_element).to eq '1470s'
+    end
+
+    it 'adds A.D. to early years' do
+      smods_rec.from_str <<-XML
+        #{mods_origin_info_start_str}
+          <dateIssued encoding="edtf">988</dateIssued>
+        #{mods_origin_info_end_str}
+      XML
+
+      imp = stanford_mods_imprint(smods_rec)
+      updated_element = imp.send(:date_str)
+      expect(updated_element).to eq '988 A.D.'
+    end
+
+    it 'adds B.C. to B.C. years' do
+      smods_rec.from_str <<-XML
+        #{mods_origin_info_start_str}
+          <dateIssued encoding="edtf">-5</dateIssued>
+        #{mods_origin_info_end_str}
+      XML
+
+      imp = stanford_mods_imprint(smods_rec)
+      updated_element = imp.send(:date_str)
+      expect(updated_element).to eq '6 B.C.'
+    end
+
+    it 'has special handling for the year 0 (1 B.C.)' do
+      smods_rec.from_str <<-XML
+        #{mods_origin_info_start_str}
+          <dateIssued>0</dateIssued>
+        #{mods_origin_info_end_str}
+      XML
+
+      imp = stanford_mods_imprint(smods_rec)
+      updated_element = imp.send(:date_str)
+      expect(updated_element).to eq '1 B.C.'
+    end
+
+    it 'presents years + months' do
+      smods_rec.from_str <<-XML
+        #{mods_origin_info_start_str}
+          <dateIssued encoding="edtf">1948-04</dateIssued>
+        #{mods_origin_info_end_str}
+      XML
+
+      imp = stanford_mods_imprint(smods_rec)
+      updated_element = imp.send(:date_str)
+      expect(updated_element).to eq 'April 1948'
+    end
+
+    it 'presents exact days nicely' do
+      smods_rec.from_str <<-XML
+        #{mods_origin_info_start_str}
+          <dateIssued encoding="edtf">1948-04-02</dateIssued>
+        #{mods_origin_info_end_str}
+      XML
+
+      imp = stanford_mods_imprint(smods_rec)
+      updated_element = imp.send(:date_str)
+      expect(updated_element).to eq 'April  2, 1948'
+    end
+
+    it 'handles very precise EDTF ranges' do
+      smods_rec.from_str <<-XML
+        #{mods_origin_info_start_str}
+          <dateIssued encoding="edtf">2014-01/2020-12-31</dateIssued>
+        #{mods_origin_info_end_str}
+      XML
+
+      imp = stanford_mods_imprint(smods_rec)
+      updated_element = imp.send(:date_str)
+      expect(updated_element).to eq 'January 2014 - December 31, 2020'
+    end
+
+    xit 'handles BC EDTF centuries' do
+      # ruby-edtf apparently can't handle this format
+      smods_rec.from_str <<-XML
+        #{mods_origin_info_start_str}
+          <dateIssued encoding="edtf">-09XX</dateIssued>
+        #{mods_origin_info_end_str}
+      XML
+
+      imp = stanford_mods_imprint(smods_rec)
+      updated_element = imp.send(:date_str)
+      expect(updated_element).to eq '10th century B.C.'
+    end
+
+    it 'handles the approximate qualifier' do
+      smods_rec.from_str <<-XML
+        #{mods_origin_info_start_str}
+          <dateIssued encoding="edtf" qualifier="approximate">1948-04-02</dateIssued>
+        #{mods_origin_info_end_str}
+      XML
+
+      imp = stanford_mods_imprint(smods_rec)
+      updated_element = imp.send(:date_str)
+      expect(updated_element).to eq '[ca. April  2, 1948]'
+    end
+
+    it 'handles the questionable qualifier' do
+      smods_rec.from_str <<-XML
+        #{mods_origin_info_start_str}
+          <dateIssued encoding="edtf" qualifier="questionable">0322</dateIssued>
+        #{mods_origin_info_end_str}
+      XML
+
+      imp = stanford_mods_imprint(smods_rec)
+      updated_element = imp.send(:date_str)
+      expect(updated_element).to eq '[322 A.D.?]'
+    end
+
+    it 'handles the inferred qualifier' do
+      smods_rec.from_str <<-XML
+        #{mods_origin_info_start_str}
+          <dateIssued encoding="edtf" qualifier="inferred">190X</dateIssued>
+        #{mods_origin_info_end_str}
+      XML
+
+      imp = stanford_mods_imprint(smods_rec)
+      updated_element = imp.send(:date_str)
+      expect(updated_element).to eq '[1900s]'
+    end
+
+    it 'presents date ranges with matching qualifiers by compacting the qualifiers' do
+      smods_rec.from_str <<-XML
+        #{mods_origin_info_start_str}
+          <dateIssued point="start" encoding="edtf" qualifier="approximate">1874</dateIssued>
+          <dateIssued point="end" encoding="w3cdtf" qualifier="approximate">1876</dateIssued>
+        #{mods_origin_info_end_str}
+      XML
+
+      imp = stanford_mods_imprint(smods_rec)
+      updated_element = imp.send(:date_str)
+      expect(updated_element).to eq '[ca. 1874 - 1876]'
+    end
+
+
+    it 'presents date ranges with mismatching qualifiers by presenting them individually' do
+      smods_rec.from_str <<-XML
+        #{mods_origin_info_start_str}
+          <dateIssued point="start" encoding="edtf" qualifier="approximate">1874</dateIssued>
+          <dateIssued point="end" encoding="w3cdtf" qualifier="questionable">1876</dateIssued>
+        #{mods_origin_info_end_str}
+      XML
+
+      imp = stanford_mods_imprint(smods_rec)
+      updated_element = imp.send(:date_str)
+      expect(updated_element).to eq '[ca. 1874] - [1876?]'
     end
   end
 
@@ -231,8 +287,8 @@ describe Stanford::Mods::Imprint do
           <publisher>Chronicle Books,</publisher>
           <dateIssued>2015.</dateIssued>' +
         mods_origin_info_end_str)
-      imprint_strs = stanford_mods_imprint(smods_rec).imprint_statements
-      expect(imprint_strs).to eq ['San Francisco : Chronicle Books, 2015.']
+      imprint_strs = stanford_mods_imprint(smods_rec).display_str
+      expect(imprint_strs).to eq 'San Francisco : Chronicle Books, 2015.'
     end
   end
 
@@ -246,8 +302,8 @@ describe Stanford::Mods::Imprint do
            <placeTerm type="text" authority="marccountry">[Amsterdam]</placeTerm>
          </place>' +
         mods_origin_info_end_str)
-      imprint_strs = stanford_mods_imprint(smods_rec).imprint_statements
-      expect(imprint_strs).to eq ['[Amsterdam]']
+      imprint_strs = stanford_mods_imprint(smods_rec).display_str
+      expect(imprint_strs).to eq '[Amsterdam]'
     end
     it "translates encoded place if there isn't a text (or non-typed) value available" do
       smods_rec.from_str(mods_origin_info_start_str +
@@ -255,8 +311,8 @@ describe Stanford::Mods::Imprint do
           <placeTerm type="code" authority="marccountry">ne</placeTerm>
         </place>' +
         mods_origin_info_end_str)
-      imprint_strs = stanford_mods_imprint(smods_rec).imprint_statements
-      expect(imprint_strs).to eq ['Netherlands']
+      imprint_strs = stanford_mods_imprint(smods_rec).display_str
+      expect(imprint_strs).to eq 'Netherlands'
     end
     it "ignores 'xx' country code" do
       smods_rec.from_str(mods_origin_info_start_str +
@@ -265,8 +321,8 @@ describe Stanford::Mods::Imprint do
          </place>
          <dateIssued>1994</dateIssued>' +
         mods_origin_info_end_str)
-      imprint_strs = stanford_mods_imprint(smods_rec).imprint_statements
-      expect(imprint_strs).to eq(['1994'])
+      imprint_strs = stanford_mods_imprint(smods_rec).display_str
+      expect(imprint_strs).to eq('1994')
     end
   end
 end

data/spec/lib/stanford-mods/coordinate_spec.rb

@@ -9,10 +9,6 @@ describe Stanford::Mods::Coordinate do
       expect(described_class.new('W80°--E100°/N487°--S42°')).not_to be_valid
     end
 
-    it 'rejects coordinates without degree symbols' do
-      expect(described_class.new('W 650--W 100/N 700--N 550')).not_to be_valid
-    end
-
     it 'rejects malformed coordinates' do
       expect(described_class.new('(E29°--E35/°S12°--S16°).')).not_to be_valid
     end
@@ -43,7 +39,9 @@ describe Stanford::Mods::Coordinate do
       %((W 170⁰--E 55⁰/N 40⁰--S 36⁰).) =>
         '-170.0 -36.0 55.0 40.0', # superscript 0 is almost a degree character..
       %((W 0°-W 0°/S 90°---S 90°)) =>
-        '-0.0 -90.0 -0.0 -90.0' # one dash, two dashes, three dashes.. what's the difference?
+        '-0.0 -90.0 -0.0 -90.0', # one dash, two dashes, three dashes.. what's the difference?
+      %((W 030.6--E 068.1/N 041.7--S 042.4)) =>
+        '-30.6 -42.4 68.1 41.7'
     }.each do |value, expected|
       describe 'parsing' do
         let(:subject) { described_class.new(value) }