Merge pull request #640 from sul-dlss/iss637-add-bcp47-language-tag-t…

…o-files

add BCP 47 language tag attribute for files

.rubocop.yml

@@ -116,6 +116,9 @@ Naming/PredicateName:
 
 # ----- RSpec ------
 
+RSpec/NestedGroups:
+  Max: 5
+
 RSpec/BeEq: # new in 2.9.0
   Enabled: true
 

.rubocop_todo.yml

@@ -1,6 +1,6 @@
 # This configuration was generated by
 # `rubocop --auto-gen-config --auto-gen-only-exclude`
-# on 2023-06-20 05:46:00 UTC using RuboCop version 1.52.1.
+# on 2023-11-01 19:51:40 UTC using RuboCop version 1.57.2.
 # 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
@@ -57,18 +57,11 @@ Metrics/ParameterLists:
 RSpec/DescribeClass:
   Enabled: false
 
-# Offense count: 87
+# Offense count: 224
 # Configuration parameters: CountAsOne.
 RSpec/ExampleLength:
   Max: 103
 
-# Offense count: 10
-# Configuration parameters: Max, AllowedGroups.
-RSpec/NestedGroups:
-  Exclude:
-    - 'spec/cocina/models/mapping/normalizers/mods/origin_info_normalizer_spec.rb'
-    - 'spec/cocina/models/validators/date_time_validator_spec.rb'
-
 # Offense count: 19
 RSpec/PendingWithoutReason:
   Exclude:
@@ -99,7 +92,7 @@ Style/MultilineBlockChain:
     - 'lib/cocina/models/mapping/to_mods/form.rb'
     - 'lib/cocina/models/mapping/to_mods/subject.rb'
 
-# Offense count: 239
+# Offense count: 249
 # This cop supports safe autocorrection (--autocorrect).
 # Configuration parameters: AllowHeredoc, AllowURI, URISchemes, IgnoreCopDirectives, AllowedPatterns.
 # URISchemes: http, https

Gemfile.lock

@@ -8,6 +8,7 @@ PATH
       dry-types (~> 1.1)
       edtf
       equivalent-xml
+      i18n
       jsonpath
       nokogiri
       openapi3_parser

cocina-models.gemspec

@@ -30,6 +30,7 @@ Gem::Specification.new do |spec|
   spec.add_dependency 'dry-types', '~> 1.1'
   spec.add_dependency 'edtf' # used for date/time validation
   spec.add_dependency 'equivalent-xml' # for diffing MODS
+  spec.add_dependency 'i18n' # for validating BCP 47 language tags, according to RFC 4646
   spec.add_dependency 'jsonpath' # used for date/time validation
   spec.add_dependency 'nokogiri'
   spec.add_dependency 'openapi3_parser' # Parsing openapi doc

lib/cocina/models/file.rb

@@ -22,6 +22,8 @@ class File < Struct
       attribute :version, Types::Strict::Integer
       # MIME Type of the File.
       attribute? :hasMimeType, Types::Strict::String
+      # BCP 47 language tag: https://www.rfc-editor.org/rfc/rfc4646.txt -- other applications (like media players) expect language codes of this format, see e.g. https://videojs.com/guides/text-tracks/#srclang
+      attribute? :languageTag, Types::Strict::String.optional
       # Use for the File.
       attribute? :use, Types::Strict::String
       attribute :hasMessageDigests, Types::Strict::Array.of(MessageDigest).default([].freeze)

lib/cocina/models/request_file.rb

@@ -13,6 +13,7 @@ class RequestFile < Struct
       attribute? :size, Types::Strict::Integer
       attribute :version, Types::Strict::Integer
       attribute? :hasMimeType, Types::Strict::String
+      attribute? :languageTag, Types::Strict::String.optional
       attribute? :externalIdentifier, Types::Strict::String
       attribute? :use, Types::Strict::String
       attribute :hasMessageDigests, Types::Strict::Array.of(MessageDigest).default([].freeze)

lib/cocina/models/validators/language_tag_validator.rb

@@ -0,0 +1,48 @@
+# frozen_string_literal: true
+
+module Cocina
+  module Models
+    module Validators
+      # Validates that a languageTag is valid according to RFC 4646, if one is present
+      class LanguageTagValidator
+        def self.validate(clazz, attributes)
+          new(clazz, attributes).validate
+        end
+
+        def initialize(clazz, attributes)
+          @clazz = clazz
+          @attributes = attributes
+        end
+
+        def validate
+          return unless meets_preconditions?
+
+          return if valid_language_tag?
+
+          raise ValidationError, 'The provided language tag is not valid according to RFC 4646: ' \
+                                 "#{attributes[:languageTag]}"
+        end
+
+        private
+
+        attr_reader :clazz, :attributes
+
+        def meets_preconditions?
+          file? && attributes[:languageTag].present?
+        end
+
+        def file?
+          (clazz::TYPES & File::TYPES).any?
+        rescue NameError
+          false
+        end
+
+        def valid_language_tag?
+          parsed_tag = I18n::Locale::Tag::Rfc4646.tag(attributes[:languageTag])
+
+          parsed_tag.present? && parsed_tag.is_a?(I18n::Locale::Tag::Rfc4646)
+        end
+      end
+    end
+  end
+end

lib/cocina/models/validators/validator.rb

@@ -15,7 +15,8 @@ class Validator
           # See also spec/cocina/models/validatable_spec.rb:59
           # DescriptionTypesValidator,
           DescriptionValuesValidator,
-          DateTimeValidator
+          DateTimeValidator,
+          LanguageTagValidator
         ].freeze
 
         def self.validate(clazz, attributes)

openapi.yml

@@ -1044,6 +1044,9 @@ components:
         hasMimeType:
           description: MIME Type of the File.
           type: string
+        languageTag:
+          description: "BCP 47 language tag: https://www.rfc-editor.org/rfc/rfc4646.txt -- other applications (like media players) expect language codes of this format, see e.g. https://videojs.com/guides/text-tracks/#srclang"
+          type: string
         use:
           description: Use for the File.
           type: string
@@ -1775,6 +1778,8 @@ components:
           type: integer
         hasMimeType:
           type: string
+        languageTag:
+          type: string
         externalIdentifier:
           type: string
         use:

spec/cocina/models/validators/language_tag_validator_spec.rb

@@ -0,0 +1,117 @@
+# frozen_string_literal: true
+
+require 'spec_helper'
+
+RSpec.describe Cocina::Models::Validators::LanguageTagValidator do
+  let(:validate) { described_class.validate(clazz, props) }
+
+  let(:props) { file_props }
+
+  let(:file_props) do
+    {
+      externalIdentifier: 'bc123df4567_1',
+      label: 'Page 1',
+      type: Cocina::Models::ObjectType.file,
+      version: 1,
+      access: { view: 'dark', download: 'none' },
+      administrative: {
+        publish: false,
+        shelve: false,
+        sdrPreserve: true
+      },
+      hasMessageDigests: [],
+      hasMimeType: 'text/plain',
+      filename: 'page1.txt'
+    }
+  end
+
+  context 'with no value for languageTag specified' do
+    context 'with a File' do
+      let(:clazz) { Cocina::Models::File }
+
+      it 'does not raise' do
+        expect { validate }.not_to raise_error
+      end
+    end
+
+    context 'with a RequestFile' do
+      let(:clazz) { Cocina::Models::RequestFile }
+
+      it 'does not raise' do
+        expect { validate }.not_to raise_error
+      end
+    end
+  end
+
+  context 'with RFC 4646 conformant value for languageTag specified' do
+    let(:props) do
+      file_props.dup.tap do |props|
+        props[:languageTag] = language_tag
+      end
+    end
+
+    context 'with a recognized language, script, and region' do
+      let(:language_tag) { 'ru-Cyrl-RU' }
+
+      context 'with a File' do
+        let(:clazz) { Cocina::Models::File }
+
+        it 'does not raise' do
+          expect { validate }.not_to raise_error
+        end
+      end
+
+      context 'with a RequestFile' do
+        let(:clazz) { Cocina::Models::RequestFile }
+
+        it 'does not raise' do
+          expect { validate }.not_to raise_error
+        end
+      end
+    end
+
+    context 'with an unrecognized language, script, and region' do
+      let(:language_tag) { 'foo-Barr-BZ' } # still conforms to the expected format of BCP 47/RFC 4646, should parse
+
+      context 'with a File' do
+        let(:clazz) { Cocina::Models::File }
+
+        it 'does not raise' do
+          expect { validate }.not_to raise_error
+        end
+      end
+
+      context 'with a RequestFile' do
+        let(:clazz) { Cocina::Models::RequestFile }
+
+        it 'does not raise' do
+          expect { validate }.not_to raise_error
+        end
+      end
+    end
+  end
+
+  context 'with value for languageTag specified that does not conform to RFC 4646' do
+    let(:props) do
+      file_props.dup.tap do |props|
+        props[:languageTag] = 'fooooooooooooo'
+      end
+    end
+
+    context 'with a File' do
+      let(:clazz) { Cocina::Models::File }
+
+      it 'raises a validation error' do
+        expect { validate }.to raise_error(Cocina::Models::ValidationError, 'The provided language tag is not valid according to RFC 4646: fooooooooooooo')
+      end
+    end
+
+    context 'with a RequestFile' do
+      let(:clazz) { Cocina::Models::RequestFile }
+
+      it 'raises a validation error' do
+        expect { validate }.to raise_error(Cocina::Models::ValidationError, 'The provided language tag is not valid according to RFC 4646: fooooooooooooo')
+      end
+    end
+  end
+end