Merge pull request #1227 from samvera/full_text_utf8

Update dependencies which saves and retreives charset

app/services/hyrax/persist_directly_contained_output_file_service.rb

@@ -5,22 +5,37 @@ class PersistDirectlyContainedOutputFileService < Hydra::Derivatives::PersistBas
     # This method conforms to the signature of the .call method on Hydra::Derivatives::PersistOutputFileService
     # * Persists the file within the DirectContainer specified by :container
     #
-    # @param [String] the data to be persisted
-    # @param [Hash] directives directions which can be used to determine where to persist to.
+    # @param content [String] the data to be persisted
+    # @param directives [Hash] directions which can be used to determine where to persist to.
     # @option directives [String] url URI for the parent object.
     # @option directives [String] container Name of the container association.
-    def self.call(string, directives)
-      file = Hydra::Derivatives::IoDecorator.new(string, new_mime_type(directives.fetch(:format)))
-      o_name = determine_original_name(file)
-      m_type = determine_mime_type(file)
-      uri = URI(directives.fetch(:url))
-      raise ArgumentError, "#{uri} is not an http(s) uri" unless uri.is_a?(URI::HTTP)
-      file_set = ActiveFedora::Base.find(ActiveFedora::Base.uri_to_id(uri.to_s))
-      remote_file = file_set.send("build_#{directives.fetch(:container)}".to_sym)
+    def self.call(content, directives)
+      file = io(content, directives)
+      file_set = retrieve_file_set(directives)
+      remote_file = retrieve_remote_file(file_set, directives)
       remote_file.content = file
-      remote_file.mime_type = m_type
-      remote_file.original_name = o_name
+      remote_file.mime_type = determine_mime_type(file)
+      remote_file.original_name = determine_original_name(file)
       file_set.save
     end
+
+    # @param directives [Hash] directions which can be used to determine where to persist to.
+    # @option directives [String] url URI for the parent object.
+    def self.retrieve_file_set(directives)
+      uri = URI(directives.fetch(:url))
+      raise ArgumentError, "#{uri} is not an http(s) uri" unless uri.is_a?(URI::HTTP)
+      ActiveFedora::Base.find(ActiveFedora::Base.uri_to_id(uri.to_s))
+    end
+    private_class_method :retrieve_file_set
+
+    # Override this implementation if you need a remote file from a different location
+    # @param file_set [FileSet] the container of the remote file
+    # @param directives [Hash] directions which can be used to determine where to persist to
+    # @option directives [String] container Name of the container association.
+    # @return [ActiveFedora::File]
+    def self.retrieve_remote_file(file_set, directives)
+      file_set.send("build_#{directives.fetch(:container)}".to_sym)
+    end
+    private_class_method :retrieve_remote_file
   end
 end

hyrax.gemspec

@@ -34,6 +34,7 @@ EOF
   spec.add_dependency 'hydra-head', '>= 10.5.0'
   spec.add_dependency 'hydra-editor', '~> 3.3'
   spec.add_dependency 'hydra-works', '~> 0.16'
+  spec.add_dependency 'hydra-derivatives', '~> 3.3'
   spec.add_dependency 'browse-everything', '>= 0.10.5'
   spec.add_dependency 'blacklight', '~> 6.9'
   spec.add_dependency 'blacklight-gallery', '~> 0.7'
@@ -69,7 +70,7 @@ EOF
   spec.add_dependency 'dry-struct', '~> 0.1'
   spec.add_dependency 'redlock', '>= 0.1.2'
   spec.add_dependency 'retriable', '>= 2.9', '< 4.0'
-  spec.add_dependency 'active-fedora', '>= 11.1.3'
+  spec.add_dependency 'active-fedora', '>= 11.3.1'
   spec.add_dependency 'linkeddata' # Required for getting values from geonames
 
   spec.add_development_dependency 'engine_cart', '~> 1.0'

spec/services/hyrax/persist_directly_contained_output_file_service_spec.rb

@@ -1,16 +1,18 @@
 RSpec.describe Hyrax::PersistDirectlyContainedOutputFileService do
   # PersistDirectlyContainedOutputFileService is used by FullTextExtract.output_file_service
-  let(:object) { FileSet.create! { |fs| fs.apply_depositor_metadata('justin') } }
-  let(:stream) { "fake file content" }
+  let(:file_set) { FileSet.create! { |fs| fs.apply_depositor_metadata('justin') } }
+  let(:content) { "fake file content" }
   subject(:call) do
-    described_class.call(stream,
+    described_class.call(content,
                          format: 'txt',
-                         url: object.uri,
+                         url: file_set.uri,
                          container: 'extracted_text')
   end
+  let(:resource) { file_set.reload.extracted_text }
 
   it "persists the file to the specified destination on the given object" do
     expect(call).to be true
-    expect(object.reload.extracted_text.content).to eq("fake file content")
+    expect(resource.content).to eq("fake file content")
+    expect(resource.content.encoding).to eq(Encoding::UTF_8)
   end
 end