assembly-objectfile 1.10.1 → 1.12.0

checksums.yaml

@@ -1,7 +1,7 @@
 ---
 SHA256:
-  metadata.gz: bfe0d362cd776d4099d57530dc6c502e8760eb49953dfa93d755016b98c54d6a
-  data.tar.gz: c1f0da284aa258aa89375b6d7b92dd7847b3e87b27712c2802e29157b86b6d9c
+  metadata.gz: e964d66148904c190829abd7d9c65d09459405f184a48a2350602d1f4c6b6a5f
+  data.tar.gz: 85fca1341b29f0c3f5b3858b419e0273754de1a47140b5ae36d2cbf2a77feea4
 SHA512:
-  metadata.gz: dfce4baa7f4d0cd52944195661ba94b6af1010fce66d714eb7c7b54778681fd5041d92d5f2fb82561c20048f76e4c313050afd880d7eb8b0c6eafb0fd6aedb67
-  data.tar.gz: c936f71be1c89887f76a9e3c66a37b690730f40a4bcc01605dcd2610d961f85d44eb110b91f40754346ef46482ad481a1aa6a76a2d9e866de42782cb344115b3
+  metadata.gz: f2cf693279f7c43392f92b3dd8bf28df4baca673276c3c96208b420c6ec26415b37b9498149a0ee6fe8ff4606046768582107f1e71b66199f04463b8f5dec918
+  data.tar.gz: fcdc3f9041410416893e8ba1258c677a9eb0f6ede6eb833deb74f9dcb30e005d226d4486419a6fc2a1ae24b9e0ceda579e1acb7f189cf022af822fc32e275fa6

data/.github/pull_request_template.md

@@ -1,12 +1,10 @@
-## Why was this change made?
+## Why was this change made? 🤔
 
 
 
-## How was this change tested?
+## How was this change tested? 🤨
 
-
-
-## Which documentation and/or configurations were updated?
+⚡ ⚠ If this change has cross service impact, ***run [integration tests](https://github.com/sul-dlss/infrastructure-integration-test) that do accessioning*** and/or test in [stage|qa] environment, in addition to specs. ⚡
 
 
 

data/.rubocop_todo.yml

@@ -1,33 +1,11 @@
 # This configuration was generated by
 # `rubocop --auto-gen-config`
-# on 2021-01-08 23:41:37 UTC using RuboCop version 1.8.0.
+# on 2021-04-26 18:43:37 UTC using RuboCop version 1.11.0.
 # The point is for the user to remove these configuration records
 # one by one as the offenses are removed from the code base.
 # Note that changes in the inspected code, or installation of new
 # versions of RuboCop, may require this file to be generated again.
 
-# Offense count: 1
-# Cop supports --auto-correct.
-# Configuration parameters: EnforcedStyleAlignWith, Severity.
-# SupportedStylesAlignWith: start_of_line, begin
-Layout/BeginEndAlignment:
-  Exclude:
-    - 'lib/assembly-objectfile/object_fileable.rb'
-
-# Offense count: 1
-# Cop supports --auto-correct.
-# Configuration parameters: AllowAliasSyntax, AllowedMethods.
-# AllowedMethods: alias_method, public, protected, private
-Layout/EmptyLinesAroundAttributeAccessor:
-  Exclude:
-    - 'lib/assembly-objectfile/object_fileable.rb'
-
-# Offense count: 1
-# Cop supports --auto-correct.
-Layout/RescueEnsureAlignment:
-  Exclude:
-    - 'lib/assembly-objectfile/object_fileable.rb'
-
 # Offense count: 1
 Lint/UselessAssignment:
   Exclude:
@@ -36,38 +14,38 @@ Lint/UselessAssignment:
 # Offense count: 3
 # Configuration parameters: IgnoredMethods, CountRepeatedAttributes.
 Metrics/AbcSize:
-  Max: 52
+  Max: 55
 
 # Offense count: 1
 # Configuration parameters: CountComments, CountAsOne, ExcludedMethods, IgnoredMethods.
 # IgnoredMethods: refine
 Metrics/BlockLength:
-  Max: 26
+  Max: 27
 
 # Offense count: 2
 # Configuration parameters: IgnoredMethods.
 Metrics/CyclomaticComplexity:
-  Max: 13
+  Max: 14
 
 # Offense count: 4
 # Configuration parameters: CountComments, CountAsOne, ExcludedMethods, IgnoredMethods.
 Metrics/MethodLength:
-  Max: 30
+  Max: 31
 
 # Offense count: 1
 # Configuration parameters: CountComments, CountAsOne.
 Metrics/ModuleLength:
-  Max: 117
+  Max: 120
 
 # Offense count: 1
 # Configuration parameters: CountKeywordArgs, MaxOptionalParameters.
 Metrics/ParameterLists:
-  Max: 11
+  Max: 12
 
 # Offense count: 2
 # Configuration parameters: IgnoredMethods.
 Metrics/PerceivedComplexity:
-  Max: 14
+  Max: 15
 
 # Offense count: 1
 # Configuration parameters: ExpectMatchingDefinition, CheckDefinitionPathHierarchy, Regex, IgnoreExecutableScripts, AllowedAcronyms.
@@ -87,7 +65,7 @@ Naming/PredicateName:
     - 'spec/**/*'
     - 'lib/assembly-objectfile/object_fileable.rb'
 
-# Offense count: 8
+# Offense count: 9
 # Configuration parameters: Max.
 RSpec/ExampleLength:
   Exclude:
@@ -101,13 +79,13 @@ RSpec/FilePath:
     - 'spec/content_metadata_spec.rb'
     - 'spec/object_file_spec.rb'
 
-# Offense count: 71
+# Offense count: 74
 # Configuration parameters: AssignmentOnly.
 RSpec/InstanceVariable:
   Exclude:
     - 'spec/object_file_spec.rb'
 
-# Offense count: 38
+# Offense count: 40
 RSpec/MultipleExpectations:
   Max: 29
 
@@ -132,29 +110,7 @@ Style/CommentedKeyword:
   Exclude:
     - 'lib/assembly-objectfile/content_metadata.rb'
 
-# Offense count: 2
-# Cop supports --auto-correct.
-Style/KeywordParametersOrder:
-  Exclude:
-    - 'lib/assembly-objectfile/content_metadata/file.rb'
-    - 'lib/assembly-objectfile/content_metadata/file_set.rb'
-
-# Offense count: 1
-# Cop supports --auto-correct.
-Style/RedundantAssignment:
-  Exclude:
-    - 'lib/assembly-objectfile/content_metadata.rb'
-
-# Offense count: 5
-# Cop supports --auto-correct.
-Style/StringConcatenation:
-  Exclude:
-    - 'config/boot.rb'
-    - 'lib/assembly-objectfile.rb'
-    - 'lib/assembly-objectfile/object_file.rb'
-    - 'spec/spec_helper.rb'
-
-# Offense count: 113
+# Offense count: 122
 # Cop supports --auto-correct.
 # Configuration parameters: AutoCorrect, AllowHeredoc, AllowURI, URISchemes, IgnoreCopDirectives, IgnoredPatterns.
 # URISchemes: http, https

data/Gemfile

@@ -4,3 +4,5 @@ source 'http://rubygems.org'
 
 # Specify your gem's dependencies in .gemspec
 gemspec
+
+gem 'byebug'

data/assembly-objectfile.gemspec

@@ -20,7 +20,7 @@ Gem::Specification.new do |s|
   s.executables   = s.files.grep(%r{^exe/}) { |f| File.basename(f) }
   s.require_paths = ['lib']
 
-  s.required_ruby_version = '~> 2.5'
+  s.required_ruby_version = '>= 2.5', '< 4'
 
   s.add_dependency 'activesupport', '>= 5.2.0'
   s.add_dependency 'deprecation'

data/lib/assembly-objectfile/content_metadata/config.rb

@@ -13,12 +13,14 @@ module Assembly
     # Represents a configuration for generating the content metadata
     class Config < Dry::Struct
       STYLES = %w[image file book map 3d document webarchive-seed].freeze
+      READING_ORDERS = %w[ltr rtl].freeze
       attribute :auto_labels, Types::Strict::Bool.default(true)
       attribute :flatten_folder_structure, Types::Strict::Bool.default(false)
       attribute :add_file_attributes, Types::Strict::Bool.default(false)
       attribute :add_exif, Types::Strict::Bool.default(false)
       attribute :file_attributes, Types::Strict::Hash.default({}.freeze)
       attribute :type, Types::Strict::String.enum(*STYLES)
+      attribute :reading_order, Types::Strict::String.default('ltr').enum(*READING_ORDERS)
     end
   end
 end

data/lib/assembly-objectfile/content_metadata/file.rb

@@ -35,7 +35,7 @@ module Assembly
         @style = style
       end
 
-      delegate :sha1, :md5, :provider_md5, :provider_sha1, :mimetype, :filesize, :image?, to: :file
+      delegate :sha1, :md5, :provider_md5, :provider_sha1, :mimetype, :filesize, :image?, :valid_image?, to: :file
 
       def file_id(common_path:, flatten_folder_structure:)
         # set file id attribute, first check the relative_path parameter on the object, and if it is set, just use that

data/lib/assembly-objectfile/content_metadata/nokogiri_builder.rb

@@ -15,6 +15,8 @@ module Assembly
 
         Nokogiri::XML::Builder.new do |xml|
           xml.contentMetadata(objectId: druid.to_s, type: config.type) do
+            xml.bookData(readingOrder: config.reading_order) if config.type == 'book'
+
             filesets.each_with_index do |fileset, index| # iterate over all the resources
               # start a new resource element
               sequence = index + 1
@@ -29,19 +31,19 @@ module Assembly
                 resource_label = fileset.label_from_file(default: default_label)
 
                 xml.label(resource_label) unless resource_label.empty?
-                fileset.files.each do |obj| # iterate over all the files in a resource
-                  xml_file_params = { id: obj.file_id(common_path: common_path, flatten_folder_structure: config.flatten_folder_structure) }
-                  xml_file_params.merge!(obj.file_attributes(config.file_attributes)) if config.add_file_attributes
-                  xml_file_params.merge!(mimetype: obj.mimetype, size: obj.filesize) if config.add_exif
+                fileset.files.each do |cm_file| # iterate over all the files in a resource
+                  xml_file_params = { id: cm_file.file_id(common_path: common_path, flatten_folder_structure: config.flatten_folder_structure) }
+                  xml_file_params.merge!(cm_file.file_attributes(config.file_attributes)) if config.add_file_attributes
+                  xml_file_params.merge!(mimetype: cm_file.mimetype, size: cm_file.filesize) if config.add_exif
 
                   xml.file(xml_file_params) do
                     if config.add_exif # add exif info if the user requested it
-                      xml.checksum(obj.sha1, type: 'sha1')
-                      xml.checksum(obj.md5, type: 'md5')
-                      xml.imageData(obj.image_data) if obj.image? # add image data for an image
-                    elsif obj.provider_md5 || obj.provider_sha1 # if we did not add exif info, see if there are user supplied checksums to add
-                      xml.checksum(obj.provider_sha1, type: 'sha1') if obj.provider_sha1
-                      xml.checksum(obj.provider_md5, type: 'md5') if obj.provider_md5
+                      xml.checksum(cm_file.sha1, type: 'sha1')
+                      xml.checksum(cm_file.md5, type: 'md5')
+                      xml.imageData(cm_file.image_data) if cm_file.valid_image? # add image data for an image
+                    elsif cm_file.provider_md5 || cm_file.provider_sha1 # if we did not add exif info, see if there are user supplied checksums to add
+                      xml.checksum(cm_file.provider_sha1, type: 'sha1') if cm_file.provider_sha1
+                      xml.checksum(cm_file.provider_md5, type: 'md5') if cm_file.provider_md5
                     end
                   end
                 end

data/lib/assembly-objectfile/content_metadata.rb

@@ -52,13 +52,15 @@ module Assembly
     #   :flatten_folder_structure = optional - Will remove *all* folder structure when genearting file IDs (e.g. DPG subfolders like '00','05' will be removed) when generating file IDs.  This is useful if the folder structure is flattened when staging files (like for DPG).
     #                                             The default is false.  If set to true, will override the "preserve_common_paths" parameter.
     #   :auto_labels = optional - Will add automated resource labels (e.g. "File 1") when labels are not provided by the user.  The default is true.
+    #   See https://consul.stanford.edu/pages/viewpage.action?spaceKey=chimera&title=DOR+content+types%2C+resource+types+and+interpretive+metadata for next two settings
+    #   :reading_order = optional - only valid for simple_book, can be 'rtl' or 'ltr'.  The default is 'ltr'.
     # Example:
     #    Assembly::ContentMetadata.create_content_metadata(:druid=>'druid:nx288wh8889',:style=>:simple_image,:objects=>object_files,:add_file_attributes=>false)
     def self.create_content_metadata(druid:, objects:, auto_labels: true,
                                      add_exif: false, bundle: :default, style: :simple_image,
                                      add_file_attributes: false, file_attributes: {},
                                      preserve_common_paths: false, flatten_folder_structure: false,
-                                     include_root_xml: nil)
+                                     include_root_xml: nil, reading_order: 'ltr')
 
       common_path = find_common_path(objects) unless preserve_common_paths # find common paths to all files provided if needed
 
@@ -68,6 +70,7 @@ module Assembly
                           add_file_attributes: add_file_attributes,
                           file_attributes: file_attributes,
                           add_exif: add_exif,
+                          reading_order: reading_order,
                           type: object_level_type(style))
 
       builder = NokogiriBuilder.build(druid: druid,

data/lib/assembly-objectfile/object_fileable.rb

@@ -8,7 +8,7 @@ module Assembly
   module ObjectFileable
     attr_accessor :file_attributes, :label, :path, :provider_md5, :provider_sha1, :relative_path, :mime_type_order
 
-    VALID_MIMETYPE_METHODS = %i[exif file extension].freeze
+    VALID_MIMETYPE_METHODS = %i[override exif file extension].freeze
 
     # @param [String] path full path to the file to be worked with
     # @param [Hash<Symbol => Object>] params options used during content metadata generation
@@ -18,7 +18,8 @@ module Assembly
     # @option params [String] :provider_sha1 pre-computed SHA1 checksum
     # @option params [String] :relative_path if you want the file ids in the content metadata it can be set, otherwise content metadata will get the full path
     # @option params [Array] :mime_type_order can be set to the order in which you want mimetypes to be determined
-    #                                          options are :exif (from exif if exists), :extension (from file extension), and :file (from unix file system command)
+    #                                          options are :override (from manual overide mapping if exists), :exif (from exif if exists),
+    #                                                      :extension (from file extension), and :file (from unix file system command)
     #                                          the default is defined in the private `default_mime_type_order` method but you can override to set your own order
     # @example
     #   Assembly::ObjectFile.new('/input/path_to_file.tif')
@@ -127,12 +128,21 @@ module Assembly
         mimetype = ''
         mime_type_order.each do |mime_type_method|
           mimetype = public_send("#{mime_type_method}_mimetype") if VALID_MIMETYPE_METHODS.include?(mime_type_method)
-          break if !mimetype.nil? && mimetype != ''
+          break if mimetype.present?
         end
         mimetype
       end
     end
 
+    # Returns mimetype information using the manual override mapping (based on a file extension lookup)
+    # @return [String] mime type for supplied file if a mapping exists for the file's extension
+    # @example
+    #   source_file = Assembly::ObjectFile.new('/input/path_to_file.json')
+    #   puts source_file.override_mimetype # 'application/json'
+    def override_mimetype
+      @override_mimetype ||= Assembly::OVERRIDE_MIMETYPES.fetch(ext.to_sym, '')
+    end
+
     # Returns mimetype information using the mime-types gem (based on a file extension lookup)
     # @return [String] mime type for supplied file
     # @example
@@ -257,7 +267,7 @@ exif&.mimetype && prefer_exif
 
     # prive method defining default preferred ordering of how mimetypes are determined
     def default_mime_type_order
-      %i[exif file extension]
+      %i[override exif file extension]
     end
 
     # private method to check for file existence before operating on it

data/lib/assembly-objectfile/version.rb

@@ -4,6 +4,6 @@
 module Assembly
   class ObjectFile
     # Project version number
-    VERSION = '1.10.1'
+    VERSION = '1.12.0'
   end
 end

data/lib/assembly-objectfile.rb

@@ -14,6 +14,14 @@ module Assembly
   #  by the file command, then a check will be made to see if exif data exists...if so, the mimetype returned by the exif data will be used
   #  if no exif data exists, then the mimetype returned by the unix file command will be used
   TRUSTED_MIMETYPES = ['text/plain', 'plain/text', 'application/pdf', 'text/html', 'application/xml'].freeze
+
+  # this is a manual override mapping of file extension to mimetype; if a file with the given extension is found, the mapped
+  #  mimetype will be returned and no further methods will be used - this is used to force a specific mimetype to be returned for
+  #  for a given file extension regardless of what exif or the unix file system command returns
+  #  the mapping format is "extension with period: returned mimetype",  e.g. for any .json file, you will always get `application/json`
+  OVERRIDE_MIMETYPES = {
+    '.json': 'application/json'
+  }.freeze
 end
 
 require 'assembly-objectfile/content_metadata'

data/spec/content_metadata_spec.rb

@@ -8,8 +8,8 @@ RSpec.describe Assembly::ContentMetadata do
 
     let(:xml) { Nokogiri::XML(result) }
 
-    context 'when style is simple_image' do
-      context 'when using a single tif and jp2' do
+    context 'when style=simple_image' do
+      context 'when using a single tif and jp2 with add_exif: true' do
         it 'generates valid content metadata with exif, adding file attributes' do
           objects = [Assembly::ObjectFile.new(TEST_TIF_INPUT_FILE), Assembly::ObjectFile.new(TEST_JP2_INPUT_FILE)]
           result = described_class.create_content_metadata(druid: TEST_DRUID, add_exif: true, add_file_attributes: true, objects: objects)
@@ -44,46 +44,8 @@ RSpec.describe Assembly::ContentMetadata do
           expect(xml.xpath('//resource/file/imageData')[1].attributes['width'].value).to eq('100')
           expect(xml.xpath('//resource/file/imageData')[1].attributes['height'].value).to eq('100')
         end
-      end
-
-      context 'when using a single tif and jp2' do
-        it 'generates valid content metadata with no exif adding specific file attributes for 2 objects, and defaults for 1 object' do
-          obj1 = Assembly::ObjectFile.new(TEST_TIF_INPUT_FILE)
-          obj2 = Assembly::ObjectFile.new(TEST_JP2_INPUT_FILE)
-          obj3 = Assembly::ObjectFile.new(TEST_JP2_INPUT_FILE2)
-          obj1.file_attributes = { publish: 'no', preserve: 'no', shelve: 'no' }
-          obj2.file_attributes = { publish: 'yes', preserve: 'yes', shelve: 'yes' }
-          objects = [obj1, obj2, obj3]
-          result = described_class.create_content_metadata(druid: TEST_DRUID, add_exif: false, add_file_attributes: true, objects: objects)
-          expect(result.class).to be String
-          xml = Nokogiri::XML(result)
-          expect(xml.errors.size).to eq 0
-          expect(xml.xpath('//contentMetadata')[0].attributes['type'].value).to eq('image')
-          expect(xml.xpath('//resource').length).to eq 3
-          expect(xml.xpath('//resource/file').length).to eq 3
-          expect(xml.xpath('//resource/file/checksum').length).to eq 0
-          expect(xml.xpath('//resource/file/imageData').length).to eq 0
-          expect(xml.xpath('//label').length).to eq 3
-          expect(xml.xpath('//label')[0].text).to match(/Image 1/)
-          expect(xml.xpath('//label')[1].text).to match(/Image 2/)
-          expect(xml.xpath('//label')[2].text).to match(/Image 3/)
-          expect(xml.xpath('//resource')[0].attributes['type'].value).to eq('image')
-          expect(xml.xpath('//resource')[1].attributes['type'].value).to eq('image')
-          expect(xml.xpath('//resource')[2].attributes['type'].value).to eq('image')
-          expect(xml.xpath('//resource/file')[0].attributes['publish'].value).to eq('no') # specificially set in object
-          expect(xml.xpath('//resource/file')[0].attributes['preserve'].value).to eq('no') # specificially set in object
-          expect(xml.xpath('//resource/file')[0].attributes['shelve'].value).to eq('no') # specificially set in object
-          expect(xml.xpath('//resource/file')[1].attributes['publish'].value).to eq('yes') # specificially set in object
-          expect(xml.xpath('//resource/file')[1].attributes['preserve'].value).to eq('yes') # specificially set in object
-          expect(xml.xpath('//resource/file')[1].attributes['shelve'].value).to eq('yes')  # specificially set in object
-          expect(xml.xpath('//resource/file')[2].attributes['publish'].value).to eq('yes') # defaults by mimetype
-          expect(xml.xpath('//resource/file')[2].attributes['preserve'].value).to eq('no') # defaults by mimetype
-          expect(xml.xpath('//resource/file')[2].attributes['shelve'].value).to eq('yes') # defaults by mimetype
-        end
-      end
 
-      context 'when using a single tif and jp2' do
-        it 'generates valid content metadata with exif, overriding file labels' do
+        it 'generates valid content metadata, overriding file labels' do
           objects = [Assembly::ObjectFile.new(TEST_TIF_INPUT_FILE, label: 'Sample tif label!'), Assembly::ObjectFile.new(TEST_JP2_INPUT_FILE, label: 'Sample jp2 label!')]
           result = described_class.create_content_metadata(druid: TEST_DRUID, add_exif: true, add_file_attributes: true, objects: objects)
           expect(result.class).to be String
@@ -119,6 +81,42 @@ RSpec.describe Assembly::ContentMetadata do
         end
       end
 
+      context 'when using a single tif and jp2 with add_exif: false' do
+        it 'generates valid content metadata adding specific file attributes for 2 objects, and defaults for 1 object' do
+          obj1 = Assembly::ObjectFile.new(TEST_TIF_INPUT_FILE)
+          obj2 = Assembly::ObjectFile.new(TEST_JP2_INPUT_FILE)
+          obj3 = Assembly::ObjectFile.new(TEST_JP2_INPUT_FILE2)
+          obj1.file_attributes = { publish: 'no', preserve: 'no', shelve: 'no' }
+          obj2.file_attributes = { publish: 'yes', preserve: 'yes', shelve: 'yes' }
+          objects = [obj1, obj2, obj3]
+          result = described_class.create_content_metadata(druid: TEST_DRUID, add_exif: false, add_file_attributes: true, objects: objects)
+          expect(result.class).to be String
+          xml = Nokogiri::XML(result)
+          expect(xml.errors.size).to eq 0
+          expect(xml.xpath('//contentMetadata')[0].attributes['type'].value).to eq('image')
+          expect(xml.xpath('//resource').length).to eq 3
+          expect(xml.xpath('//resource/file').length).to eq 3
+          expect(xml.xpath('//resource/file/checksum').length).to eq 0
+          expect(xml.xpath('//resource/file/imageData').length).to eq 0
+          expect(xml.xpath('//label').length).to eq 3
+          expect(xml.xpath('//label')[0].text).to match(/Image 1/)
+          expect(xml.xpath('//label')[1].text).to match(/Image 2/)
+          expect(xml.xpath('//label')[2].text).to match(/Image 3/)
+          expect(xml.xpath('//resource')[0].attributes['type'].value).to eq('image')
+          expect(xml.xpath('//resource')[1].attributes['type'].value).to eq('image')
+          expect(xml.xpath('//resource')[2].attributes['type'].value).to eq('image')
+          expect(xml.xpath('//resource/file')[0].attributes['publish'].value).to eq('no') # specificially set in object
+          expect(xml.xpath('//resource/file')[0].attributes['preserve'].value).to eq('no') # specificially set in object
+          expect(xml.xpath('//resource/file')[0].attributes['shelve'].value).to eq('no') # specificially set in object
+          expect(xml.xpath('//resource/file')[1].attributes['publish'].value).to eq('yes') # specificially set in object
+          expect(xml.xpath('//resource/file')[1].attributes['preserve'].value).to eq('yes') # specificially set in object
+          expect(xml.xpath('//resource/file')[1].attributes['shelve'].value).to eq('yes')  # specificially set in object
+          expect(xml.xpath('//resource/file')[2].attributes['publish'].value).to eq('yes') # defaults by mimetype
+          expect(xml.xpath('//resource/file')[2].attributes['preserve'].value).to eq('no') # defaults by mimetype
+          expect(xml.xpath('//resource/file')[2].attributes['shelve'].value).to eq('yes') # defaults by mimetype
+        end
+      end
+
       context 'when using a single tif and jp2' do
         it 'generates valid content metadata with exif, overriding file labels for one, and skipping auto labels for the others or for where the label is set but is blank' do
           objects = [Assembly::ObjectFile.new(TEST_TIF_INPUT_FILE, label: 'Sample tif label!'), Assembly::ObjectFile.new(TEST_JP2_INPUT_FILE), Assembly::ObjectFile.new(TEST_JP2_INPUT_FILE, label: '')]
@@ -331,9 +329,29 @@ RSpec.describe Assembly::ContentMetadata do
           end
         end
       end
+
+      context 'when using a single svg with add_exif: true' do
+        subject(:result) { described_class.create_content_metadata(druid: TEST_DRUID, add_exif: true, auto_labels: false, add_file_attributes: true, objects: objects) }
+
+        let(:objects) { [Assembly::ObjectFile.new(TEST_SVG_INPUT_FILE)] }
+
+        it 'generates no imageData node' do
+          xml = Nokogiri::XML(result)
+          expect(xml.errors.size).to eq 0
+          expect(xml.xpath('//contentMetadata')[0].attributes['type'].value).to eq('image')
+          expect(xml.xpath('//resource/file').length).to eq 1
+          expect(xml.xpath('//resource/file')[0]['mimetype']).to eq 'image/svg+xml'
+          expect(xml.xpath('//resource/file')[0]['publish']).to eq('no')
+          expect(xml.xpath('//resource/file')[0]['preserve']).to eq('yes')
+          expect(xml.xpath('//resource/file')[0]['shelve']).to eq('no')
+          expect(xml.xpath("//resource[@sequence='1']/file").length).to eq 1
+          expect(xml.xpath('//imageData')).not_to be_present
+          expect(xml.xpath('//resource')[0].attributes['type'].value).to eq('image')
+        end
+      end
     end
 
-    context 'when style is webarchive-seed' do
+    context 'when style=webarchive-seed' do
       context 'when using a jp2' do
         it 'generates valid content metadata with exif, adding file attributes' do
           objects = [Assembly::ObjectFile.new(TEST_JP2_INPUT_FILE)]
@@ -342,6 +360,7 @@ RSpec.describe Assembly::ContentMetadata do
           xml = Nokogiri::XML(result)
           expect(xml.errors.size).to eq 0
           expect(xml.xpath('//contentMetadata')[0].attributes['type'].value).to eq('webarchive-seed')
+          expect(xml.xpath('//bookData').length).to eq 0
           expect(xml.xpath('//resource').length).to eq 1
           expect(xml.xpath('//resource/file').length).to eq 1
           expect(xml.xpath('//resource/file/checksum').length).to eq 2
@@ -375,6 +394,7 @@ RSpec.describe Assembly::ContentMetadata do
           xml = Nokogiri::XML(result)
           expect(xml.errors.size).to eq 0
           expect(xml.xpath('//contentMetadata')[0].attributes['type'].value).to eq('map')
+          expect(xml.xpath('//bookData').length).to eq 0
           expect(xml.xpath('//resource/file').length).to eq 2
           expect(xml.xpath('//resource/file')[0].attributes['mimetype']).to be_nil
           expect(xml.xpath('//resource/file')[0].attributes['publish'].value).to eq('yes')
@@ -404,6 +424,7 @@ RSpec.describe Assembly::ContentMetadata do
           expect(xml.errors.size).to eq 0
           expect(xml.xpath('//contentMetadata')[0].attributes['type'].value).to eq('book')
           expect(xml.xpath('//contentMetadata')[0].attributes['objectId'].value).to eq(TEST_DRUID.to_s)
+          expect(xml.xpath('//bookData')[0].attributes['readingOrder'].value).to eq('ltr')
           expect(xml.xpath('//resource').length).to eq 4
           expect(xml.xpath('//resource/file').length).to eq 6
           expect(xml.xpath("//resource[@sequence='1']/file")[0].attributes['id'].value).to eq('oo000oo0001_00_001.tif')
@@ -428,20 +449,21 @@ RSpec.describe Assembly::ContentMetadata do
         end
       end
 
-      context "when item has a 'druid:' prefix. Using two tifs, two associated jp2s, two associated pdfs and one lingering PDF using bundle=dpg" do
+      context "when item has a 'druid:' prefix and specified book order. Using two tifs, two associated jp2s, two associated pdfs and one lingering PDF using bundle=dpg" do
         it 'generates valid content metadata with flattening folder structure' do
           objects = [Assembly::ObjectFile.new(TEST_DPG_TIF), Assembly::ObjectFile.new(TEST_DPG_JP),
                      Assembly::ObjectFile.new(TEST_DPG_PDF), Assembly::ObjectFile.new(TEST_DPG_TIF2),
                      Assembly::ObjectFile.new(TEST_DPG_JP2), Assembly::ObjectFile.new(TEST_DPG_PDF2),
                      Assembly::ObjectFile.new(TEST_DPG_SPECIAL_PDF1)]
           test_druid = "druid:#{TEST_DRUID}"
-          result = described_class.create_content_metadata(druid: test_druid, bundle: :dpg, objects: objects, style: :simple_book, flatten_folder_structure: true)
+          result = described_class.create_content_metadata(druid: test_druid, bundle: :dpg, objects: objects, style: :simple_book, flatten_folder_structure: true, reading_order: 'rtl')
           expect(result.class).to be String
           expect(result.include?('<?xml')).to be true
           xml = Nokogiri::XML(result)
           expect(xml.errors.size).to eq 0
           expect(xml.xpath('//contentMetadata')[0].attributes['type'].value).to eq('book')
           expect(xml.xpath('//contentMetadata')[0].attributes['objectId'].value).to eq(test_druid)
+          expect(xml.xpath('//bookData')[0].attributes['readingOrder'].value).to eq('rtl')
           expect(test_druid).to eq("druid:#{TEST_DRUID}")
           expect(xml.xpath('//resource').length).to eq 3
           expect(xml.xpath('//resource/file').length).to be 7
@@ -466,6 +488,14 @@ RSpec.describe Assembly::ContentMetadata do
         end
       end
 
+      context 'throws an error with invalid reading order' do
+        subject(:result) { described_class.create_content_metadata(druid: "druid:#{TEST_DRUID}", bundle: :dpg, objects: [], style: :simple_book, flatten_folder_structure: true, reading_order: 'bogus') }
+
+        it 'generates valid content metadata with flattening folder structure' do
+          expect { result }.to raise_error(Dry::Struct::Error)
+        end
+      end
+
       context 'when using two tifs' do
         it 'generates valid content metadata for two tifs of style=simple_book' do
           objects = [Assembly::ObjectFile.new(TEST_TIF_INPUT_FILE), Assembly::ObjectFile.new(TEST_TIF_INPUT_FILE2)]
@@ -493,7 +523,7 @@ RSpec.describe Assembly::ContentMetadata do
       end
     end
 
-    context 'when using style=book_with_pdf' do
+    context 'when style=book_with_pdf' do
       context 'when using two tiffs and a pdf' do
         let(:objects) do
           [Assembly::ObjectFile.new(TEST_TIF_INPUT_FILE),
@@ -580,6 +610,7 @@ RSpec.describe Assembly::ContentMetadata do
           xml = Nokogiri::XML(result)
           expect(xml.errors.size).to eq 0
           expect(xml.xpath('//contentMetadata')[0].attributes['type'].value).to eq('file')
+          expect(xml.xpath('//bookData').length).to eq 0
           expect(xml.xpath('//resource').length).to eq 4
           expect(xml.xpath('//resource/file').length).to eq 4
           expect(xml.xpath('//label').length).to eq 4
@@ -611,6 +642,7 @@ RSpec.describe Assembly::ContentMetadata do
       it 'generates valid content metadata for two tifs' do
         expect(xml.errors.size).to eq 0
         expect(xml.xpath('//contentMetadata')[0].attributes['type'].value).to eq('book')
+        expect(xml.xpath('//bookData').length).to eq 1
         expect(xml.xpath('//resource').length).to eq 2
         expect(xml.xpath('//resource/file').length).to eq 2
         expect(xml.xpath('//label').length).to eq 2
@@ -641,6 +673,7 @@ RSpec.describe Assembly::ContentMetadata do
       it 'generates valid content metadata' do
         expect(xml.errors.size).to eq 0
         expect(xml.xpath('//contentMetadata')[0].attributes['type'].value).to eq('document')
+        expect(xml.xpath('//bookData').length).to eq 0
         expect(xml.xpath('//resource').length).to eq 1
         expect(xml.xpath('//resource/file').length).to eq 1
         expect(xml.xpath('//label').length).to eq 1
@@ -713,6 +746,7 @@ RSpec.describe Assembly::ContentMetadata do
       it 'generates valid content metadata' do
         expect(xml.errors.size).to eq 0
         expect(xml.xpath('//contentMetadata')[0].attributes['type'].value).to eq('3d')
+        expect(xml.xpath('//bookData').length).to eq 0
         expect(xml.xpath('//resource').length).to eq 4
         expect(xml.xpath('//resource/file').length).to eq 4
         expect(xml.xpath('//label').length).to eq 4

data/spec/object_file_spec.rb

@@ -70,6 +70,13 @@ describe Assembly::ObjectFile do
     expect(@ai.mimetype).to eq('text/plain')
   end
 
+  it 'overrides the mimetype generators and uses the manual mapping to set the correct mimetype of application/json for a .json file' do
+    @ai = described_class.new(TEST_JSON_FILE)
+    expect(@ai.exif_mimetype).to be_nil # exif returns nil
+    expect(@ai.file_mimetype).to eq('text/plain') # unix file system command returns plain text
+    expect(@ai.mimetype).to eq('application/json') # but our configured mapping overrides both and returns application/json
+  end
+
   it 'sets the correct mimetype of image/tiff for .tif files' do
     @ai = described_class.new(TEST_TIF_INPUT_FILE)
     expect(@ai.mimetype).to eq('image/tiff')

data/spec/spec_helper.rb

@@ -3,8 +3,8 @@
 require 'simplecov'
 SimpleCov.start
 
-bootfile = File.expand_path("#{File.dirname(__FILE__)}/../config/boot")
-require bootfile
+require File.expand_path("#{File.dirname(__FILE__)}/../config/boot")
+require 'byebug'
 
 RSpec.configure do |config|
   config.order = 'random'
@@ -18,6 +18,7 @@ TEST_TIF_INPUT_FILE2  = File.join(TEST_INPUT_DIR, 'test2.tif')
 TEST_JPEG_INPUT_FILE  = File.join(TEST_INPUT_DIR, 'test.jpg')
 TEST_JP2_INPUT_FILE = File.join(TEST_INPUT_DIR, 'test.jp2')
 TEST_JP2_INPUT_FILE2 = File.join(TEST_INPUT_DIR, 'test2.jp2')
+TEST_SVG_INPUT_FILE  = File.join(TEST_INPUT_DIR, 'test.svg')
 TEST_JP2_OUTPUT_FILE = File.join(TEST_OUTPUT_DIR, 'test.jp2')
 TEST_PDF_FILE = File.join(TEST_INPUT_DIR, 'test.pdf')
 
@@ -54,6 +55,8 @@ TEST_RES3_TEI = File.join(TEST_INPUT_DIR, 'res3_teifile.txt')
 
 TEST_FILE_NO_EXIF = File.join(TEST_INPUT_DIR, 'file_with_no_exif.xml')
 
+TEST_JSON_FILE = File.join(TEST_INPUT_DIR, 'test.json')
+
 TEST_OBJ_FILE = File.join(TEST_INPUT_DIR, 'someobject.obj')
 TEST_PLY_FILE = File.join(TEST_INPUT_DIR, 'someobject.ply')
 

data/spec/test_data/input/test.json

@@ -0,0 +1 @@
+{some: 'thing'}

data/spec/test_data/input/test.svg

@@ -0,0 +1,2 @@
+<?xml version="1.0" encoding="UTF-8"?>
+<svg width="507.3" height="508.2" xmlns="http://www.w3.org/2000/svg" xmlns:xlink="http://www.w3.org/1999/xlink" overflow="hidden"><defs><clipPath id="clip0"><rect x="2361" y="213" width="507" height="508"/></clipPath></defs><g clip-path="url(#clip0)" transform="translate(-2361 -213)"><path d="M2361 467C2361 326.72 2474.5 213 2614.5 213 2754.5 213 2868 326.72 2868 467 2868 607.28 2754.5 721 2614.5 721 2474.5 721 2361 607.28 2361 467Z" fill="#BF9000" fill-rule="evenodd"/></g></svg>

metadata

@@ -1,7 +1,7 @@
 --- !ruby/object:Gem::Specification
 name: assembly-objectfile
 version: !ruby/object:Gem::Version
-  version: 1.10.1
+  version: 1.12.0
 platform: ruby
 authors:
 - Peter Mangiafico
@@ -11,7 +11,7 @@ authors:
 autorequire:
 bindir: exe
 cert_chain: []
-date: 2021-03-16 00:00:00.000000000 Z
+date: 2022-02-18 00:00:00.000000000 Z
 dependencies:
 - !ruby/object:Gem::Dependency
   name: activesupport
@@ -263,7 +263,9 @@ files:
 - spec/test_data/input/someobject.obj
 - spec/test_data/input/someobject.ply
 - spec/test_data/input/test.jp2
+- spec/test_data/input/test.json
 - spec/test_data/input/test.pdf
+- spec/test_data/input/test.svg
 - spec/test_data/input/test.tif
 - spec/test_data/input/test2.jp2
 - spec/test_data/input/test2.tif
@@ -278,16 +280,19 @@ require_paths:
 - lib
 required_ruby_version: !ruby/object:Gem::Requirement
   requirements:
-  - - "~>"
+  - - ">="
     - !ruby/object:Gem::Version
       version: '2.5'
+  - - "<"
+    - !ruby/object:Gem::Version
+      version: '4'
 required_rubygems_version: !ruby/object:Gem::Requirement
   requirements:
   - - ">="
     - !ruby/object:Gem::Version
       version: '0'
 requirements: []
-rubygems_version: 3.2.14
+rubygems_version: 3.2.32
 signing_key:
 specification_version: 4
 summary: Ruby immplementation of file services needed to prepare objects to be accessioned
@@ -326,7 +331,9 @@ test_files:
 - spec/test_data/input/someobject.obj
 - spec/test_data/input/someobject.ply
 - spec/test_data/input/test.jp2
+- spec/test_data/input/test.json
 - spec/test_data/input/test.pdf
+- spec/test_data/input/test.svg
 - spec/test_data/input/test.tif
 - spec/test_data/input/test2.jp2
 - spec/test_data/input/test2.tif