Add support for a @ROLE attribute on a contentMetadata's file element

lib/dor/datastreams/content_metadata_ds.rb

@@ -19,6 +19,7 @@ class ContentMetadataDS < ActiveFedora::OmDatastream
           t.mimeType :path => { :attribute => 'mimeType' }, :index_as => [:displayable]
           t.dataType :path => { :attribute => 'dataType' }, :index_as => [:displayable]
           t.size     :path => { :attribute => 'size'     }, :index_as => [:displayable]    # , :data_type => :long
+          t.role     :path => { :attribute => 'role' },     :index_as => [:not_searchable]
           t.shelve   :path => { :attribute => 'shelve'   }, :index_as => [:not_searchable] # , :data_type => :boolean
           t.publish  :path => { :attribute => 'publish'  }, :index_as => [:not_searchable] # , :data_type => :boolean
           t.preserve :path => { :attribute => 'preserve' }, :index_as => [:not_searchable] # , :data_type => :boolean
@@ -85,6 +86,7 @@ def to_solr(solr_doc = {}, *args)
       shelved_size = 0
       counts = Hash.new(0)                # default count is zero
       resource_type_counts = Hash.new(0)  # default count is zero
+      file_roles = ::Set.new
       mime_types = ::Set.new
       first_shelved_image = nil
 
@@ -100,6 +102,7 @@ def to_solr(solr_doc = {}, *args)
             first_shelved_image ||= file['id'] if file['id'] =~ /jp2$/
           end
           mime_types << file['mimetype']
+          file_roles << file['role'] if file['role']
         end
       end
       solr_doc['content_type_ssim'              ] = doc.root['type']
@@ -110,6 +113,7 @@ def to_solr(solr_doc = {}, *args)
       solr_doc['preserved_size_dbtsi'           ] = preserved_size # double (trie) to support very large sizes
       solr_doc['shelved_size_dbtsi'             ] = shelved_size # double (trie) to support very large sizes
       solr_doc['resource_types_ssim'            ] = resource_type_counts.keys if resource_type_counts.size > 0
+      solr_doc['content_file_roles_ssim'        ] = file_roles.to_a if file_roles.size > 0
       resource_type_counts.each do |key, count|
         solr_doc["#{key}_resource_count_itsi"] = count
       end
@@ -146,6 +150,7 @@ def add_file(file, resource_name)
       end
       file_node['size'    ] = file[:size     ] if file[:size     ]
       file_node['mimetype'] = file[:mime_type] if file[:mime_type]
+      file_node['role'] = file[:role] if file[:role]
       file_node
     end
 
@@ -207,6 +212,7 @@ def add_resource(files, resource_name, position, type = 'file')
           file_node.add_child(checksum_node)
         }
         file_node['size'] = file[:size] if file[:size]
+        file_node['role'] = file[:role] if file[:role]
       end
       ng_xml.search('//contentMetadata').first.add_child(node)
       node
@@ -236,12 +242,15 @@ def remove_file(file_name)
     # @param [String] publish
     # @param [String] shelve
     # @param [String] preserve
-    def update_attributes(file_name, publish, shelve, preserve)
+    def update_attributes(file_name, publish, shelve, preserve, attributes = {})
       self.ng_xml_will_change!
       file_node = ng_xml.search('//file[@id=\'' + file_name + '\']').first
       file_node['publish' ] = publish
       file_node['shelve'  ] = shelve
       file_node['preserve'] = preserve
+      attributes.each do |key, value|
+        file_node[key] = value
+      end
     end
 
     # @param file [Object] some hash-like file
@@ -261,7 +270,7 @@ def update_file(file, old_file_id)
         checksum_node.content = file[algo]
       }
 
-      [:size, :shelve, :preserve, :publish].each { |x|
+      [:size, :shelve, :preserve, :publish, :role].each { |x|
         file_node[x.to_s] = file[x] if file[x]
       }
     end

spec/datastreams/content_metadata_ds_spec.rb

@@ -14,7 +14,7 @@
     <checksum type="md5">3d3ff46d98f3d517d0bf086571e05c18</checksum>
     <checksum type="sha1">ca1eb0edd09a21f9dd9e3a89abc790daf4d04916</checksum>
     </file>
-    <file format="GIF" id="gw177fc7976_05_0001.gif" mimetype="image/gif" preserve="no" publish="no" shelve="no" size="4128877">
+    <file format="GIF" id="gw177fc7976_05_0001.gif" mimetype="image/gif" preserve="no" publish="no" shelve="no" size="4128877" role="derivative">
     <imageData height="4580" width="5939"/>
     <checksum type="md5">406d5d80fdd9ecc0352d339badb4a8fb</checksum>
     <checksum type="sha1">61940d4fad097cba98a3e9dd9f12a90dde0be1ac</checksum>
@@ -105,6 +105,18 @@
         expect(checksum.content).to eq(checksum['type'] == 'md5' ? '123456' : '56789')
       end
     end
+
+    it 'should add a file with a role="transcription"' do
+      files = [
+        @file.merge(:name => 'transcription.txt', :role => 'transcription', :size => '23456', :preserve => 'yes')
+      ]
+      @cm.add_resource(files, 'resource', 1, 'page')
+      nodes = @cm.ng_xml.search('//resource[@id=\'resource\']/file')
+      expect(nodes.length).to eq(1)
+      node = nodes.first
+      expect(node['id'      ]).to eq('transcription.txt')
+      expect(node['role'    ]).to eq('transcription')
+    end
   end
 
   describe 'remove_resource' do
@@ -128,7 +140,7 @@
   end
   describe 'add_file' do
     it 'should add a file to the resource' do
-      @cm.add_file(@file, '0001')
+      @cm.add_file(@file.merge(role: 'some-role'), '0001')
       xml = @cm.ng_xml
       hits = xml.search('//resource[@id=\'0001\']/file')
       expect(hits.length).to eq(4)
@@ -138,20 +150,22 @@
       expect(new_file['publish' ]).to eq('no')
       expect(new_file['preserve']).to eq('no')
       expect(new_file['size'    ]).to eq('12345')
+      expect(new_file['role'    ]).to eq('some-role')
       expect(@cm).to be_changed
     end
   end
 
   describe 'update_file' do
     it 'should modify an existing file record' do
-      @cm.update_file(@file, 'gw177fc7976_05_0001.jp2')
+      @cm.update_file(@file.merge(role: 'some-role'), 'gw177fc7976_05_0001.jp2')
       file = @cm.ng_xml.search('//file[@id=\'new_file.jp2\']')
       expect(file.length).to eq(1)
       file = file.first
       expect(file['shelve'  ]).to eq('no')
       expect(file['publish' ]).to eq('no')
       expect(file['preserve']).to eq('no')
       expect(file['size'    ]).to eq('12345')
+      expect(file['role'    ]).to eq('some-role')
     end
     it 'should error out if there isnt an existing record to modify' do
       expect { @cm.update_file(@file, 'gw177fc7976_05_0001_different.jp2')}.to raise_error(StandardError)
@@ -201,6 +215,7 @@
       expected = {
         'content_type_ssim'               => 'map',
         'content_file_mimetypes_ssim'     => ['image/jp2', 'image/gif', 'image/tiff'],
+        'content_file_roles_ssim'         => ['derivative'],
         'shelved_content_file_count_itsi' => 1,
         'resource_count_itsi'             => 1,
         'content_file_count_itsi'         => 3,