Merge pull request #626 from sul-dlss/escape-uri

Add support for files with spaces

lib/file_uris.rb

@@ -0,0 +1,30 @@
+# frozen_string_literal: true
+
+# Builds the list of URIs for files in an object
+class FileUris
+  # @param [String] druid
+  # @param [Cocina::Models::DRO]
+  def self.build(druid, obj)
+    filenames = extract_filenames(obj)
+
+    workspace = DruidTools::Druid.new(druid, File.absolute_path(Settings.sdr.local_workspace_root))
+    content_dir = workspace.find_filelist_parent('content', filenames)
+
+    filenames.map do |filename|
+      URI::File.build(path: File.join(content_dir, filename).gsub(' ', '%20')).to_s
+    end
+  end
+
+  def self.extract_filenames(obj)
+    filenames = []
+    obj.structural.contains.each do |fileset|
+      next if fileset.structural.contains.blank?
+
+      fileset.structural.contains.each do |file|
+        filenames << file.label
+      end
+    end
+    filenames
+  end
+  private_class_method :extract_filenames
+end

lib/robots/dor_repo/accession/technical_metadata.rb

@@ -18,41 +18,14 @@ def perform(druid)
           # skip if no files
           return LyberCore::Robot::ReturnState.new(status: :skipped, note: 'object has no files') if obj.structural.contains.blank?
 
-          file_uris = extract_file_uris(druid, obj)
+          file_uris = FileUris.build(druid, obj)
           invoke_techmd_service(druid, file_uris)
 
           LyberCore::Robot::ReturnState.new(status: :noop, note: 'Initiated technical metadata generation from technical-metadata-service.')
         end
 
         private
 
-        def object_for_druid
-          Dor::Services::Client.object(druid).find
-        end
-
-        def extract_filenames(obj)
-          filenames = []
-          obj.structural.contains.each do |fileset|
-            next if fileset.structural.contains.blank?
-
-            fileset.structural.contains.each do |file|
-              filenames << file.label
-            end
-          end
-          filenames
-        end
-
-        def extract_file_uris(druid, obj)
-          filenames = extract_filenames(obj)
-
-          workspace = DruidTools::Druid.new(druid, File.absolute_path(Settings.sdr.local_workspace_root))
-          content_dir = workspace.find_filelist_parent('content', filenames)
-
-          # In future Ruby's, this could be: filenames.map { |filename| URI::File.build(path: File.join(content_dir, filename)).to_s }
-          # However, in 2.5.3, URI::File does not exist.
-          filenames.map { |filename| "file://#{File.join(content_dir, filename)}" }
-        end
-
         def invoke_techmd_service(druid, file_uris)
           req = JSON.generate(druid: druid, files: file_uris)
           resp = Faraday.post("#{Settings.tech_md_service.url}/v1/technical-metadata", req,

spec/lib/file_uris_spec.rb

@@ -0,0 +1,52 @@
+# frozen_string_literal: true
+
+RSpec.describe FileUris do
+  let(:workspace) { File.absolute_path('spec/fixtures/workspace') }
+  let(:druid) { 'druid:dd116zh0343' }
+  let(:root) { File.absolute_path(Settings.sdr.local_workspace_root) }
+  let(:object) do
+    Cocina::Models::DRO.new(externalIdentifier: druid,
+                            type: Cocina::Models::Vocab.object,
+                            label: 'my repository object',
+                            version: 1,
+                            structural: {
+                              contains: [{
+                                externalIdentifier: '222',
+                                type: Cocina::Models::Vocab.fileset,
+                                label: 'my repository object',
+                                version: 1,
+                                structural: {
+                                  contains: [
+                                    {
+                                      externalIdentifier: '222-1',
+                                      label: filename,
+                                      type: Cocina::Models::Vocab.file,
+                                      version: 1
+                                    }
+                                  ]
+                                }
+                              }]
+                            })
+  end
+
+  before do
+    # For File URIs, need to use absolute paths
+    allow(Settings.sdr).to receive(:local_workspace_root).and_return(workspace)
+  end
+
+  describe '.build' do
+    subject { described_class.build(druid, object) }
+
+    context 'with a sub folder' do
+      let(:filename) { 'folder1PuSu/story1u.txt' }
+
+      it { is_expected.to eq ["file://#{root}/dd/116/zh/0343/dd116zh0343/content/folder1PuSu/story1u.txt"] }
+    end
+
+    context 'with a space' do
+      let(:filename) { 'file with space.txt' }
+
+      it { is_expected.to eq ["file://#{root}/dd/116/zh/0343/dd116zh0343/content/file%20with%20space.txt"] }
+    end
+  end
+end