Extract xml deserialization from the ResourceFile model

app/models/embed/purl/file_xml_deserializer.rb

@@ -0,0 +1,33 @@
+# frozen_string_literal: true
+
+module Embed
+  class Purl
+    class FileXmlDeserializer
+      # @param [String] druid the identifier of the resource this file belongs to.
+      # @param [String] description the label for this file
+      # @param [Nokogiri::XML::Element] file
+      # @param [Dor::RightsAuth] rights
+      def initialize(druid, description, file, rights)
+        @druid = druid
+        @description = description
+        @file = file
+        @rights = rights
+      end
+
+      def deserialize
+        ResourceFile.new(
+          druid: @druid,
+          label: @description,
+          filename: @file.attributes['id'].value,
+          mimetype: @file.attributes['mimetype']&.value,
+          size: @file.attributes['size']&.value.to_i,
+          rights: @rights
+        ) do |file|
+          if @file.xpath('./*/@duration').present?
+            file.duration = Embed::MediaDuration.new(@file.xpath('./*[@duration]').first).to_s
+          end
+        end
+      end
+    end
+  end
+end

app/models/embed/purl/resource.rb

@@ -29,7 +29,7 @@ def three_dimensional?
       # @return [Array<ResourceFile>]
       def files
         @files ||= @resource.xpath('./file').map do |file|
-          ResourceFile.new(druid, description, file, @rights)
+          FileXmlDeserializer.new(druid, description, file, @rights).deserialize
         end
       end
 

app/models/embed/purl/resource_file.rb

@@ -3,24 +3,20 @@
 module Embed
   class Purl
     class ResourceFile
-      # @param [String] druid the identifier of the resource this file belongs to.
-      # @param [String] description the label for this file
-      # @param [Nokogiri::XML::Element] file
-      # @param [Dor::RightsAuth] rights
-      def initialize(druid, description, file, rights)
-        @druid = druid
-        @description = description
-        @file = file
-        @rights = rights
+      def initialize(attributes = {})
+        self.attributes = attributes
+        yield(self) if block_given?
       end
 
-      def label
-        @description
+      def attributes=(hash)
+        hash.each do |key, value|
+          public_send("#{key}=", value)
+        end
       end
 
-      def title
-        @file.attributes['id'].try(:value)
-      end
+      attr_accessor :druid, :label, :filename, :mimetype, :size, :duration, :rights
+
+      alias title filename
 
       ##
       # Creates a file url for stacks
@@ -50,24 +46,10 @@ def pdf?
         mimetype == 'application/pdf'
       end
 
-      def mimetype
-        @file.attributes['mimetype'].try(:value)
-      end
-
       def image?
         mimetype =~ %r{image/jp2}i
       end
 
-      # @return [Integer]
-      def size
-        @file.attributes['size']&.value.to_i
-      end
-
-      def duration
-        md = Embed::MediaDuration.new(@file.xpath('./*[@duration]').first) if @file.xpath('./*/@duration').present?
-        md&.to_s
-      end
-
       def stanford_only?
         value, _rule = @rights.stanford_only_rights_for_file(title)
 
@@ -79,7 +61,7 @@ def location_restricted?
       end
 
       def world_downloadable?
-        @rights.world_downloadable_file?(@file.attributes['id'])
+        @rights.world_downloadable_file?(title)
       end
     end
   end

spec/models/embed/purl/file_xml_deserializer_spec.rb

@@ -0,0 +1,76 @@
+# frozen_string_literal: true
+
+require 'rails_helper'
+
+RSpec.describe Embed::Purl::FileXmlDeserializer do
+  describe '#deserialize' do
+    let(:file_node) do
+      Nokogiri::XML(
+        <<~XML
+          <file size="12345" mimetype="image/jp2" id="myvideo.mp4"/>
+        XML
+      ).root
+    end
+
+    let(:resource_file) { described_class.new('abc123', 'desc', file_node, nil).deserialize }
+
+    it 'creates a resource file' do
+      expect(resource_file.druid).to eq 'abc123'
+      expect(resource_file.label).to eq 'desc'
+      expect(resource_file.mimetype).to eq 'image/jp2'
+      expect(resource_file.size).to eq 12_345
+      expect(resource_file.filename).to eq 'myvideo.mp4'
+      expect(resource_file.duration).to be_nil
+    end
+
+    context 'with duration in videoData' do
+      let(:file_node) do
+        Nokogiri::XML(
+          <<~XML
+            <file size="12345" mimetype="image/jp2" id="myvideo.mp4">
+              <videoData duration='P0DT1H2M3S'/>
+            </file>
+          XML
+        ).root
+      end
+
+      it 'gets the duration string' do
+        expect(resource_file.duration).to eq '1:02:03'
+      end
+    end
+
+    context 'with duration in audioData' do
+      let(:file_node) do
+        Nokogiri::XML(
+          <<~XML
+            <file size="12345" mimetype="image/jp2" id="myvideo.mp4">
+              <audioData duration='PT43S'/>
+            </file>
+          XML
+        ).root
+      end
+
+      it 'gets the duration string' do
+        expect(resource_file.duration).to eq '0:43'
+      end
+    end
+
+    context 'when the duration in audioData is invalid' do
+      let(:file_node) do
+        Nokogiri::XML(
+          <<~XML
+            <file size="12345" mimetype="image/jp2" id="myvideo.mp4">
+              <audioData duration='invalid'/>
+            </file>
+          XML
+        ).root
+      end
+
+      it 'returns nil and logs an error' do
+        allow(Honeybadger).to receive(:notify)
+        expect(resource_file.duration).to be_nil
+        expect(Honeybadger).to have_received(:notify).with("ResourceFile#media duration ISO8601::Errors::UnknownPattern: 'invalid'")
+      end
+    end
+  end
+end

spec/models/embed/purl/resource_file_spec.rb

@@ -27,17 +27,9 @@
   end
 
   describe '#file_url' do
-    let(:file_node) do
-      Nokogiri::XML(<<~XML
-        <file size="12345" mimetype="image/jp2" id="#{file_name}"/>
-      XML
-                   ).root
-    end
-
-    let(:file_name) { 'cool_file' }
-
+    let(:filename) { 'cool_file' }
     let(:resource) { instance_double(Embed::Purl::Resource, druid: 'abc123') }
-    let(:resource_file) { described_class.new('abc123', 'desc', file_node, nil) }
+    let(:resource_file) { described_class.new(druid: 'abc123', filename:) }
 
     it 'creates a stacks file url' do
       expect(resource_file.file_url).to eq 'https://stacks.stanford.edu/file/druid:abc123/cool_file'
@@ -48,15 +40,15 @@
     end
 
     context 'when there are special characters in the file name' do
-      let(:file_name) { '[Dissertation] micro-TEC vfinal (for submission)-augmented.pdf' }
+      let(:filename) { '[Dissertation] micro-TEC vfinal (for submission)-augmented.pdf' }
 
       it 'escapes them' do
         expect(resource_file.file_url).to eq 'https://stacks.stanford.edu/file/druid:abc123/%5BDissertation%5D%20micro-TEC%20vfinal%20%28for%20submission%29-augmented.pdf'
       end
     end
 
     context 'when there are literal slashes in the file name' do
-      let(:file_name) { 'path/to/[Dissertation] micro-TEC vfinal (for submission)-augmented.pdf' }
+      let(:filename) { 'path/to/[Dissertation] micro-TEC vfinal (for submission)-augmented.pdf' }
 
       it 'allows them' do
         expect(resource_file.file_url)
@@ -76,11 +68,10 @@
   end
 
   describe '#label' do
-    let(:resource_file) { instance_double(Nokogiri::XML::Element, attributes: { 'id' => double(value: 'The File ID') }) }
+    let(:resource_file) { described_class.new(label: 'The Resource Description') }
 
     it 'is the resource description' do
-      file = described_class.new('abc123', 'The Resource Description', resource_file, double('rights'))
-      expect(file.label).to eq 'The Resource Description'
+      expect(resource_file.label).to eq 'The Resource Description'
     end
   end
 
@@ -162,42 +153,6 @@
     end
   end
 
-  describe 'duration' do
-    let(:rf) { described_class.new('123', 'desc', f, double('Rights')) }
-    let(:f) { double('File') }
-
-    it 'gets duration string from videoData' do
-      video_data_el = Nokogiri::XML("<videoData duration='P0DT1H2M3S'/>").root
-      expect(f).to receive(:xpath).with('./*/@duration').and_return(['something'])
-      expect(f).to receive(:xpath).with('./*[@duration]').and_return([video_data_el])
-      expect(Embed::MediaDuration).to receive(:new).and_call_original
-      expect(rf.duration).to eq '1:02:03'
-    end
-
-    it 'gets duration string from audioData' do
-      audio_data_el = Nokogiri::XML("<audioData duration='PT43S'/>").root
-      expect(f).to receive(:xpath).with('./*/@duration').and_return(['something'])
-      expect(f).to receive(:xpath).with('./*[@duration]').and_return([audio_data_el])
-      expect(Embed::MediaDuration).to receive(:new).and_call_original
-      expect(rf.duration).to eq '0:43'
-    end
-
-    it 'nil when missing media data element' do
-      allow(f).to receive(:xpath)
-      expect(Embed::MediaDuration).not_to receive(:new)
-      expect(rf.duration).to be_nil
-    end
-
-    it 'invalid format returns nil and logs an error' do
-      audio_data_el = Nokogiri::XML("<audioData duration='invalid'/>").root
-      expect(f).to receive(:xpath).with('./*/@duration').and_return(['something'])
-      expect(f).to receive(:xpath).with('./*[@duration]').and_return([audio_data_el])
-      expect(Honeybadger).to receive(:notify).with("ResourceFile#media duration ISO8601::Errors::UnknownPattern: 'invalid'")
-      expect(Embed::MediaDuration).to receive(:new).and_call_original
-      expect(rf.duration).to be_nil
-    end
-  end
-
   describe '#vtt?' do
     subject { file.vtt? }