Merge pull request #321 from sul-dlss/href-that-are-not

Normalize the DigitalObject href value to something usable

app/models/digital_object.rb

@@ -2,7 +2,7 @@
 
 ##
 # Override Arclight::DigitalObject to add Purl URL
-# Demo data href only contains the ID
+# Demo data href sometimes only contains the ID
 class DigitalObject
   attr_reader :label, :href
 
@@ -13,6 +13,18 @@ def initialize(label:, href:)
 
   def self.from_json(json)
     object_data = JSON.parse(json)
-    new(label: object_data['label'], href: "https://purl.stanford.edu/#{object_data['href']}")
+    new(label: object_data['label'], href: normalize_href(object_data['href']))
+  end
+
+  # Make a usable Purl URL from whatever happens to be in
+  # the DigitalObject href in the sample data.
+  # Ideally, this value would be consistent, but it is not currently.
+  def self.normalize_href(href)
+    # Some complete Purl URLs do not use https, convert them
+    return href.gsub('http://', 'https://') if href.match?(%r{https?://purl.stanford.edu})
+    # Some hrefs contain only a druid, convert them to a complete Purl URL
+    return "https://purl.stanford.edu/#{href}" if href.match?(/^([a-z]{2})(\d{3})([a-z]{2})(\d{4})$/)
+
+    href
   end
 end

spec/models/digital_object_spec.rb

@@ -4,16 +4,16 @@
 
 RSpec.describe DigitalObject do
   subject(:instance) do
-    described_class.new(label: 'An object label', href: 'an-object-id')
+    described_class.new(label: 'An object label', href: 'aa111bb2222')
   end
 
   describe 'label' do
     let(:empty_label) do
-      described_class.new(label: '', href: 'an-object-id')
+      described_class.new(label: '', href: 'aa111bb2222')
     end
 
     it 'uses href if label is blank' do
-      expect(empty_label.href).to eq 'an-object-id'
+      expect(empty_label.href).to eq 'aa111bb2222'
     end
   end
 
@@ -22,7 +22,29 @@
       deserialized = described_class.from_json(instance.to_json)
       expect(deserialized).to be_a described_class
       expect(deserialized.label).to eq 'An object label'
-      expect(deserialized.href).to eq 'https://purl.stanford.edu/an-object-id'
+      expect(deserialized.href).to eq 'https://purl.stanford.edu/aa111bb2222'
+    end
+  end
+
+  describe "#{described_class}.normalize_href" do
+    it 'returns the href unchanged if it contains something other than a druid' do
+      expect(described_class.normalize_href('some-other-id')).to eq 'some-other-id'
+    end
+
+    it 'returns the href unchanged if it contains a URL that is not a Purl' do
+      expect(described_class.normalize_href('http://www.somewebsite/some-other-id')).to eq 'http://www.somewebsite/some-other-id'
+    end
+
+    it 'returns the href unchanged if it contains a complete Purl URL' do
+      expect(described_class.normalize_href('https://purl.stanford.edu/aa111bb2222')).to eq 'https://purl.stanford.edu/aa111bb2222'
+    end
+
+    it 'returns the Purl URL but converts http to https' do
+      expect(described_class.normalize_href('http://purl.stanford.edu/aa111bb2222')).to eq 'https://purl.stanford.edu/aa111bb2222'
+    end
+
+    it 'returns a complete Purl URL if the href only contains a druid' do
+      expect(described_class.normalize_href('aa111bb2222')).to eq 'https://purl.stanford.edu/aa111bb2222'
     end
   end
 end