bulkrax 3.0.1 → 3.1.0

checksums.yaml

@@ -1,7 +1,7 @@
 ---
 SHA256:
-  metadata.gz: 7810d87388103e4ff6d160e6acb64a492dcfcf16a23f93ad6848f6cc97f2d9db
-  data.tar.gz: 75bee56959564774f20cda642b6c732cd9cb8a4fbd19222222fbac429effc949
+  metadata.gz: 23cd36b2c3dc13769f7d20da3c0d53f2d5c2cabef15d5beeeccee75aed6d679a
+  data.tar.gz: 868f0ec5c54862943e126af0ecf63d49771bc9a4ac88e566074205d7077e063c
 SHA512:
-  metadata.gz: 74297cb60a695e46fa97389b6a9a23124c816304d636267e72585a286f370c923aefe0e10727d3a215fc8d7a1eefa579cb84cd3b84b8e8f65a72619c0c10176e
-  data.tar.gz: 57e62164db12eb01bbfa200d612ecd0439099e257d760009b289644e93fd043e7edc3f95c08f023fed2746abfc2c57fac996b542abc0cf724794dc78bc422ddc
+  metadata.gz: a4f9c359a539054dba54000ca30bd1209bb862254c75463c2598807e8fd91e7e562e3bd6203f5f347ffe17da22f58d7915c1090109e88aabccac774c0281b1ee
+  data.tar.gz: 11c0083e67c4add719f04c01d20fac3216f6e994407dfdc48ae5f6d82696241a2b81ef6156279568b92e4bd56d3598085e83bac3718d04892a4fc96bbefd1028

data/app/models/bulkrax/csv_entry.rb

@@ -46,6 +46,7 @@ module Bulkrax
       add_visibility
       add_metadata_for_model
       add_rights_statement
+      sanitize_controlled_uri_values!
       add_local
 
       self.parsed_metadata

data/app/models/concerns/bulkrax/import_behavior.rb

@@ -2,7 +2,7 @@
 
 module Bulkrax
   # Import Behavior for Entry classes
-  module ImportBehavior
+  module ImportBehavior # rubocop:disable Metrics/ModuleLength
     extend ActiveSupport::Concern
 
     def build_for_importer
@@ -105,6 +105,65 @@ module Bulkrax
       end
     end
 
+    # Attempt to sanitize Questioning Authority URI values for configured controlled fields of common
+    # data entry mistakes. Controlled URI values are only valid if they are an exact match.
+    # Example:
+    #   Valid value:     http://rightsstatements.org/vocab/InC/1.0/
+    #   Provided value:  https://rightsstatements.org/vocab/InC/1.0
+    #   Sanitized value: http://rightsstatements.org/vocab/InC/1.0/ ("s" from "https" removed, trailing "/" added)
+    #
+    # @return [Boolean] true if all controlled URI values are sanitized successfully
+    def sanitize_controlled_uri_values!
+      Bulkrax.qa_controlled_properties.each do |field|
+        next if parsed_metadata[field].blank?
+
+        parsed_metadata[field].each_with_index do |value, i|
+          next if value.blank?
+
+          if (validated_uri_value = validate_value(value, field))
+            parsed_metadata[field][i] = validated_uri_value
+          else
+            debug_msg = %(Unable to locate active authority ID "#{value}" in config/authorities/#{field.pluralize}.yml)
+            Rails.logger.debug(debug_msg)
+            error_msg = %("#{value}" is not a valid and/or active authority ID for the :#{field} field)
+            raise ::StandardError, error_msg
+          end
+        end
+      end
+
+      true
+    end
+
+    # @param value [String] value to validate
+    # @param field [String] name of the controlled property
+    # @return [String, nil] validated URI value or nil
+    def validate_value(value, field)
+      if value.match?(::URI::DEFAULT_PARSER.make_regexp)
+        value = value.strip.chomp
+        # add trailing forward slash unless one is already present
+        value << '/' unless value.match?(%r{/$})
+      end
+
+      valid = if active_id_for_authority?(value, field)
+                true
+              else
+                value.include?('https') ? value.sub!('https', 'http') : value.sub!('http', 'https')
+                active_id_for_authority?(value, field)
+              end
+
+      valid ? value : nil
+    end
+
+    # @param value [String] value to check
+    # @param field [String] name of the controlled property
+    # @return [Boolean] provided value is a present, active authority ID for the provided field
+    def active_id_for_authority?(value, field)
+      field_service = ('Hyrax::' + "#{field}_service".camelcase).constantize
+      active_authority_ids = field_service.new.active_elements.map { |ae| ae['id'] }
+
+      active_authority_ids.include?(value)
+    end
+
     def factory
       @factory ||= Bulkrax::ObjectFactory.new(attributes: self.parsed_metadata,
                                               source_identifier_value: identifier,

data/lib/bulkrax/version.rb

@@ -1,5 +1,5 @@
 # frozen_string_literal: true
 
 module Bulkrax
-  VERSION = '3.0.1'
+  VERSION = '3.1.0'
 end

data/lib/bulkrax.rb

@@ -12,13 +12,13 @@ module Bulkrax
                    :related_children_field_mapping,
                    :related_parents_field_mapping,
                    :reserved_properties,
+                   :qa_controlled_properties,
                    :field_mappings,
                    :import_path,
                    :export_path,
                    :removed_image_path,
                    :server_name,
-                   :api_definition,
-                   :removed_image_path
+                   :api_definition
 
     self.parsers = [
       { name: "OAI - Dublin Core", class_name: "Bulkrax::OaiDcParser", partial: "oai_fields" },
@@ -119,6 +119,11 @@ module Bulkrax
       original_url
       relative_path
     ]
+
+    # List of Questioning Authority properties that are controlled via YAML files in
+    # the config/authorities/ directory. For example, the :rights_statement property
+    # is controlled by the active terms in config/authorities/rights_statements.yml
+    self.qa_controlled_properties = %w[rights_statement license]
   end
 
   def self.api_definition
@@ -131,8 +136,6 @@ module Bulkrax
     )
   end
 
-  self.removed_image_path = 'app/assets/images/bulkrax/removed.png'
-
   # this function maps the vars from your app into your engine
   def self.setup
     yield self

data/lib/generators/bulkrax/templates/config/initializers/bulkrax.rb

@@ -61,6 +61,12 @@ Bulkrax.setup do |config|
 
   # Properties that should not be used in imports/exports. They are reserved for use by Hyrax.
   # config.reserved_properties += ['my_field']
+
+  # List of Questioning Authority properties that are controlled via YAML files in
+  # the config/authorities/ directory. For example, the :rights_statement property
+  # is controlled by the active terms in config/authorities/rights_statements.yml
+  # Defaults: 'rights_statement' and 'license'
+  # config.qa_controlled_properties += ['my_field']
 end
 
 # Sidebar for hyrax 3+ support

metadata

@@ -1,7 +1,7 @@
 --- !ruby/object:Gem::Specification
 name: bulkrax
 version: !ruby/object:Gem::Version
-  version: 3.0.1
+  version: 3.1.0
 platform: ruby
 authors:
 - Rob Kaufman