dor-services 6.0.5 → 6.1.0

data/lib/dor/services/dublin_core_service.rb

@@ -0,0 +1,40 @@
+# frozen_string_literal: true
+
+module Dor
+  class DublinCoreService
+    MODS_TO_DC_XSLT = Nokogiri::XSLT(File.new(File.expand_path(File.dirname(__FILE__) + '/mods2dc.xslt')))
+    XMLNS_OAI_DC = 'http://www.openarchives.org/OAI/2.0/oai_dc/'
+    class CrosswalkError < RuntimeError; end
+
+    def initialize(work, include_collection_as_related_item: true)
+      @work = work
+      @include_collection = include_collection_as_related_item
+    end
+
+    # Generates Dublin Core from the MODS in the descMetadata datastream using the LoC mods2dc stylesheet
+    #    Should not be used for the Fedora DC datastream
+    # @raise [CrosswalkError] Raises an Exception if the generated DC is empty or has no children
+    # @return [Nokogiri::Doc] the DublinCore XML document object
+    def to_xml
+      dc_doc = MODS_TO_DC_XSLT.transform(desc_md)
+      dc_doc.xpath('/oai_dc:dc/*[count(text()) = 0]', oai_dc: XMLNS_OAI_DC).remove # Remove empty nodes
+      raise CrosswalkError, "Dor::Item#generate_dublin_core produced incorrect xml (no root):\n#{dc_doc.to_xml}" if dc_doc.root.nil?
+      raise CrosswalkError, "Dor::Item#generate_dublin_core produced incorrect xml (no children):\n#{dc_doc.to_xml}" if dc_doc.root.children.size == 0
+
+      dc_doc
+    end
+
+    private
+
+    def desc_md
+      return PublicDescMetadataService.new(work).ng_xml(include_access_conditions: false) if include_collection?
+
+      work.descMetadata.ng_xml
+    end
+
+    def include_collection?
+      @include_collection
+    end
+    attr_reader :work
+  end
+end

data/lib/dor/services/file_metadata_merge_service.rb

@@ -0,0 +1,67 @@
+# frozen_string_literal: true
+
+module Dor
+  # Merges contentMetadata from several objects into one.
+  class FileMetadataMergeService
+    # @param [Array<String>] secondary_druids ids of the secondary objects that will get their contentMetadata merged into this one
+    def self.copy_file_resources(primary, secondary_druids)
+      merge_service = FileMetadataMergeService.new primary, secondary_druids
+      merge_service.copy_file_resources
+    end
+
+    def initialize(primary, secondary_druids)
+      @pid = primary.pid
+      content_metadata = primary.contentMetadata
+      content_metadata.ng_xml_will_change!
+      @primary_cm = content_metadata.ng_xml
+      @secondary_druids = secondary_druids
+    end
+
+    # Appends contentMetadata file resources from the source objects to this object
+    def copy_file_resources
+      base_id = primary_cm.at_xpath('/contentMetadata/@objectId').value
+      max_sequence = primary_cm.at_xpath('/contentMetadata/resource[last()]/@sequence').value.to_i
+
+      secondary_druids.each do |src_pid|
+        source_obj = Dor.find src_pid
+        source_cm = source_obj.contentMetadata.ng_xml
+
+        # Copy the resources from each source object
+        source_cm.xpath('/contentMetadata/resource').each do |old_resource|
+          max_sequence += 1
+          resource_copy = old_resource.clone
+          resource_copy['sequence'] = max_sequence.to_s
+
+          # Append sequence number to each secondary filename, then
+          # look for filename collisions with the primary object
+          resource_copy.xpath('file').each do |secondary_file|
+            secondary_file['id'] = SecondaryFileNameService.create(secondary_file['id'], max_sequence)
+
+            if primary_cm.at_xpath("//file[@id = '#{secondary_file['id']}']")
+              raise Dor::Exception, "File '#{secondary_file['id']}' from secondary object #{src_pid} already exist in primary object: #{pid}"
+            end
+          end
+
+          if old_resource['type']
+            resource_copy['id'] = "#{old_resource['type']}_#{max_sequence}"
+          else
+            resource_copy['id'] = "#{base_id}_#{max_sequence}"
+          end
+
+          lbl = old_resource.at_xpath 'label'
+          resource_copy.at_xpath('label').content = "#{Regexp.last_match(1)} #{max_sequence}" if lbl && lbl.text =~ /^(.*)\s+\d+$/
+
+          primary_cm.at_xpath('/contentMetadata/resource[last()]').add_next_sibling resource_copy
+          attr_node = primary_cm.create_element 'attr', src_pid, name: 'mergedFromPid'
+          resource_copy.first_element_child.add_previous_sibling attr_node
+          attr_node = primary_cm.create_element 'attr', old_resource['id'], name: 'mergedFromResource'
+          resource_copy.first_element_child.add_previous_sibling attr_node
+        end
+      end
+    end
+
+    private
+
+    attr_reader :secondary_druids, :primary_cm, :pid
+  end
+end

data/lib/dor/services/indexing_service.rb

@@ -12,9 +12,13 @@ module Dor
     #   entry.  a placeholder will be used otherwise. 'request.uuid' might be useful in a Rails app.
     def self.generate_index_logger(&entry_id_block)
       index_logger = Logger.new(Config.indexing_svc.log, Config.indexing_svc.log_rotation_interval)
-      index_logger.formatter = proc do |severity, datetime, progname, msg|
+      index_logger.formatter = proc do |_severity, datetime, _progname, msg|
         date_format_str = Config.indexing_svc.log_date_format_str
-        entry_id = begin entry_id_block.call rescue '---' end
+        entry_id = begin begin
+                           entry_id_block.call
+                         rescue StandardError
+                           '---'
+                         end end
         "[#{entry_id}] [#{datetime.utc.strftime(date_format_str)}] #{msg}\n"
       end
       index_logger
@@ -49,7 +53,7 @@ module Dor
       msg = "failed to reindex #{pid}: #{e}"
       default_index_logger.error msg
       raise ReindexError.new(msg)
-    rescue => e
+    rescue StandardError => e
       default_index_logger.error "failed to reindex #{pid}: #{e}"
       raise
     end
@@ -72,7 +76,7 @@ module Dor
       options = args.pop if args.last.is_a? Hash
 
       if args.length > 0
-        warn "Dor::IndexingService.reindex_pid with primitive arguments is deprecated; pass e.g. { logger: logger, raise_errors: bool } instead"
+        warn 'Dor::IndexingService.reindex_pid with primitive arguments is deprecated; pass e.g. { logger: logger, raise_errors: bool } instead'
         index_logger, should_raise_errors = args
         index_logger ||= default_index_logger
         should_raise_errors = true if should_raise_errors.nil?

data/lib/dor/services/merge_service.rb

@@ -31,7 +31,7 @@ module Dor
     end
 
     def move_metadata_and_content
-      @primary.copy_file_resources @secondary_pids
+      FileMetadataMergeService.copy_file_resources @primary, @secondary_pids
       @primary.save
       copy_workspace_content
     end
@@ -50,9 +50,9 @@ module Dor
           primary_resource = primary_cm.at_xpath "//resource[attr[@name = 'mergedFromPid']/text() = '#{secondary.pid}' and
                                                              attr[@name = 'mergedFromResource']/text() = '#{src_resource['id']}' ]"
           sequence = primary_resource['sequence']
-          src_resource.xpath('//file/@id').map { |id| id.value }.each do |file_id|
+          src_resource.xpath('//file/@id').map(&:value).each do |file_id|
             copy_path = sec_druid.find_content file_id
-            new_name = secondary.new_secondary_file_name(file_id, sequence)
+            new_name = SecondaryFileNameService.create(file_id, sequence)
             # TODO: verify new_name exists in primary_cm?
             FileUtils.cp(copy_path, File.join(dest_path, "/#{new_name}"))
           end
@@ -71,13 +71,13 @@ module Dor
           unpublish
           Dor::CleanupService.cleanup_by_druid @current_secondary.pid
           Dor::Config.workflow.client.archive_active_workflow 'dor', @current_secondary.pid
-        rescue => e
+        rescue StandardError => e
           @logger.error "Unable to decommission #{@current_secondary.pid} with primary object #{@primary.pid}: #{e.inspect}"
           @logger.error e.backtrace.join("\n")
         end
       end
     end
-    alias_method :decomission_secondaries, :decommission_secondaries
+    alias decomission_secondaries decommission_secondaries
     deprecate decomission_secondaries: 'Use decommission_secondaries instead'
 
     # Remove content from stacks

data/lib/dor/services/metadata_handlers/catalog_handler.rb

@@ -13,7 +13,7 @@ handler = Class.new do
   def label(metadata)
     mods = Nokogiri::XML(metadata)
     mods.root.add_namespace_definition('mods', 'http://www.loc.gov/mods/v3')
-    mods.xpath('/mods:mods/mods:titleInfo[1]').xpath('mods:title|mods:nonSort').collect { |n| n.text }.join(' ').strip
+    mods.xpath('/mods:mods/mods:titleInfo[1]').xpath('mods:title|mods:nonSort').collect(&:text).join(' ').strip
   end
 
   def prefixes

data/lib/dor/services/metadata_service.rb

@@ -3,7 +3,7 @@
 require 'cache'
 
 module Dor
-  class MetadataError < Exception; end
+  class MetadataError < RuntimeError; end
 
   #  class MetadataHandler
   #
@@ -23,9 +23,7 @@ module Dor
 
       def register(handler_class)
         %w(fetch label prefixes).each do |method|
-          unless handler_class.instance_methods.include?(method) || handler_class.instance_methods.include?(method.to_sym)
-            raise TypeError, "Metadata handlers must define ##{method}"
-          end
+          raise TypeError, "Metadata handlers must define ##{method}" unless handler_class.instance_methods.include?(method) || handler_class.instance_methods.include?(method.to_sym)
         end
         handler = handler_class.new
         handler.prefixes.each do |prefix|
@@ -40,12 +38,12 @@ module Dor
 
       def can_resolve?(identifier)
         (prefix, _identifier) = identifier.split(/:/, 2)
-        handlers.keys.include?(prefix.to_sym)
+        handlers.key?(prefix.to_sym)
       end
 
       # TODO: Return a prioritized list
       def resolvable(identifiers)
-        identifiers.select { |identifier| self.can_resolve?(identifier) }
+        identifiers.select { |identifier| can_resolve?(identifier) }
       end
 
       def fetch(identifier)
@@ -78,6 +76,6 @@ module Dor
   end
 end
 
-Dir[File.join(File.dirname(__FILE__), 'metadata_handlers', '*.rb')].each { |handler_file|
+Dir[File.join(File.dirname(__FILE__), 'metadata_handlers', '*.rb')].each do |handler_file|
   load handler_file
-}
+end

data/lib/dor/services/ontology.rb

@@ -0,0 +1,35 @@
+# frozen_string_literal: true
+
+module Dor
+  class Ontology
+    extend Deprecation
+    self.deprecation_horizon = 'dor-services version 7.0.0'
+
+    def self.[](key)
+      @data[key]
+    end
+    deprecation_deprecate :[] => 'Use property() instead'
+
+    def self.key?(key)
+      @data.key?(key)
+    end
+
+    def self.include?(key)
+      @data.include?(key)
+    end
+    deprecation_deprecate include?: 'Use key? instead'
+
+    def self.property(key)
+      Term.new(@data[key])
+    end
+
+    class Term
+      def initialize(uri:, human_readable:)
+        @label = human_readable
+        @uri = uri
+      end
+
+      attr_reader :label, :uri
+    end
+  end
+end

data/lib/dor/services/open_data_license_service.rb

@@ -0,0 +1,20 @@
+# frozen_string_literal: true
+
+module Dor
+  class OpenDataLicenseService < Ontology
+    # these hashes map short ("machine") license codes to their corresponding URIs and human readable titles. they
+    # also allow for deprecated entries (via optional :deprecation_warning).  clients that use these maps are advised to
+    # only display undeprecated entries, except where a deprecated entry is already in use by an object.  e.g., an APO
+    # that already specifies "by_sa" for its default license code could continue displaying that in a list of license options
+    # for editing, preferably with the deprecation warning.  but other deprecated entries would be omitted in such a
+    # selectbox.
+    @data = {
+      'pddl' => { human_readable: 'Open Data Commons Public Domain Dedication and License 1.0',
+                  uri: 'http://opendatacommons.org/licenses/pddl/1.0/' },
+      'odc-by' => { human_readable: 'Open Data Commons Attribution License 1.0',
+                    uri: 'http://opendatacommons.org/licenses/by/1.0/' },
+      'odc-odbl' => { human_readable: 'Open Data Commons Open Database License 1.0',
+                      uri: 'http://opendatacommons.org/licenses/odbl/1.0/' }
+    }.freeze
+  end
+end

data/lib/dor/services/public_desc_metadata_service.rb

@@ -16,14 +16,21 @@ module Dor
 
     # @return [String] Public descriptive medatada XML
     def to_xml(include_access_conditions: true)
-      add_collection_reference!
-      add_access_conditions! if include_access_conditions
-      add_constituent_relations!
-      strip_comments!
-
-      new_doc = Nokogiri::XML(doc.to_xml) { |x| x.noblanks }
-      new_doc.encoding = 'UTF-8'
-      new_doc.to_xml
+      ng_xml(include_access_conditions: include_access_conditions).to_xml
+    end
+
+    # @return [Nokogiri::XML::Document]
+    def ng_xml(include_access_conditions: true)
+      @ng_xml ||= begin
+        add_collection_reference!
+        add_access_conditions! if include_access_conditions
+        add_constituent_relations!
+        strip_comments!
+
+        new_doc = Nokogiri::XML(doc.to_xml, &:noblanks)
+        new_doc.encoding = 'UTF-8'
+        new_doc
+      end
     end
 
     private
@@ -37,20 +44,20 @@ module Dor
     # @note this method modifies the passed in doc
     def add_access_conditions!
       # clear out any existing accessConditions
-      doc.xpath('//mods:accessCondition', 'mods' => 'http://www.loc.gov/mods/v3').each { |n| n.remove }
+      doc.xpath('//mods:accessCondition', 'mods' => 'http://www.loc.gov/mods/v3').each(&:remove)
       rights = object.datastreams['rightsMetadata'].ng_xml
 
       rights.xpath('//use/human[@type="useAndReproduction"]').each do |use|
         txt = use.text.strip
         next if txt.empty?
 
-        doc.root.element_children.last.add_next_sibling doc.create_element('accessCondition', txt, :type => 'useAndReproduction')
+        doc.root.element_children.last.add_next_sibling doc.create_element('accessCondition', txt, type: 'useAndReproduction')
       end
       rights.xpath('//copyright/human[@type="copyright"]').each do |cr|
         txt = cr.text.strip
         next if txt.empty?
 
-        doc.root.element_children.last.add_next_sibling doc.create_element('accessCondition', txt, :type => 'copyright')
+        doc.root.element_children.last.add_next_sibling doc.create_element('accessCondition', txt, type: 'copyright')
       end
       rights.xpath("//use/machine[#{ci_compare('type', 'creativecommons')}]").each do |lic_type|
         next if lic_type.text =~ /none/i
@@ -59,7 +66,7 @@ module Dor
         next if lic_text.empty?
 
         new_text = "CC #{lic_type.text}: #{lic_text}"
-        doc.root.element_children.last.add_next_sibling doc.create_element('accessCondition', new_text, :type => 'license')
+        doc.root.element_children.last.add_next_sibling doc.create_element('accessCondition', new_text, type: 'license')
       end
       rights.xpath("//use/machine[#{ci_compare('type', 'opendatacommons')}]").each do |lic_type|
         next if lic_type.text =~ /none/i
@@ -68,7 +75,7 @@ module Dor
         next if lic_text.empty?
 
         new_text = "ODC #{lic_type.text}: #{lic_text}"
-        doc.root.element_children.last.add_next_sibling doc.create_element('accessCondition', new_text, :type => 'license')
+        doc.root.element_children.last.add_next_sibling doc.create_element('accessCondition', new_text, type: 'license')
       end
     end
 
@@ -133,7 +140,7 @@ module Dor
       end
     end
 
-    def add_related_item_node_for_collection! collection_druid
+    def add_related_item_node_for_collection!(collection_druid)
       begin
         collection_obj = Dor.find(collection_druid)
       rescue ActiveFedora::ObjectNotFoundError

data/lib/dor/services/public_xml_service.rb

@@ -20,7 +20,7 @@ module Dor
 
       pub.add_child(public_relationships.root) unless public_relationships.nil? # TODO: Should never be nil in practice; working around an ActiveFedora quirk for testing
       pub.add_child(object.generate_dublin_core.root)
-      pub.add_child(Nokogiri::XML(object.generate_public_desc_md).root)
+      pub.add_child(PublicDescMetadataService.new(object).ng_xml.root)
       pub.add_child(release_xml.root) unless release_xml.xpath('//release').children.size == 0 # If there are no release_tags, this prevents an empty <releaseData/> from being added
       # Note we cannot base this on if an individual object has release tags or not, because the collection may cause one to be generated for an item,
       # so we need to calculate it and then look at the final result.
@@ -28,7 +28,7 @@ module Dor
       thumb = ThumbnailService.new(object).thumb
       pub.add_child(Nokogiri("<thumb>#{thumb}</thumb>").root) unless thumb.nil?
 
-      new_pub = Nokogiri::XML(pub.to_xml) { |x| x.noblanks }
+      new_pub = Nokogiri::XML(pub.to_xml, &:noblanks)
       new_pub.encoding = 'UTF-8'
       new_pub.to_xml
     end
@@ -38,11 +38,11 @@ module Dor
     def release_xml
       @release_xml ||= begin
         builder = Nokogiri::XML::Builder.new do |xml|
-          xml.releaseData {
+          xml.releaseData do
             object.released_for.each do |project, released_value|
-              xml.release(released_value['release'], :to => project)
+              xml.release(released_value['release'], to: project)
             end
-          }
+          end
         end
         Nokogiri::XML(builder.to_xml)
       end
@@ -83,7 +83,7 @@ module Dor
           src_resource_id = externalFile['resourceId']
           src_druid = externalFile['objectId']
           src_file_id = externalFile['fileId']
-          fail ArgumentError, "Malformed externalFile data: #{externalFile.inspect}" if [src_resource_id, src_file_id, src_druid].map(&:blank?).any?
+          raise ArgumentError, "Malformed externalFile data: #{externalFile.inspect}" if [src_resource_id, src_file_id, src_druid].map(&:blank?).any?
 
           # grab source item
           src_item = Dor.find(src_druid)

data/lib/dor/services/publish_metadata_service.rb

@@ -0,0 +1,100 @@
+# frozen_string_literal: true
+
+module Dor
+  # Merges contentMetadata from several objects into one.
+  class PublishMetadataService
+    # @param [Dor::Item] item the object to be publshed
+    def self.publish(item)
+      new(item).publish
+    end
+
+    def initialize(item)
+      @item = item
+    end
+
+    # Appends contentMetadata file resources from the source objects to this object
+    def publish
+      return unpublish unless world_discoverable?
+
+      transfer_metadata
+      publish_notify_on_success
+    end
+
+    private
+
+    attr_reader :item
+
+    def transfer_metadata
+      dc_xml = item.generate_dublin_core.to_xml(&:no_declaration)
+      transfer_to_document_store(dc_xml, 'dc')
+      %w(identityMetadata contentMetadata rightsMetadata).each do |stream|
+        transfer_to_document_store(item.datastreams[stream].content.to_s, stream) if item.datastreams[stream]
+      end
+      transfer_to_document_store(PublicXmlService.new(item).to_xml, 'public')
+      transfer_to_document_store(PublicDescMetadataService.new(self).to_xml, 'mods')
+    end
+
+    # Clear out the document cache for this item
+    def unpublish
+      purl_druid.prune!
+      publish_delete_on_success
+    end
+
+    def world_discoverable?
+      rights = item.rightsMetadata.ng_xml.clone.remove_namespaces!
+      rights.at_xpath("//rightsMetadata/access[@type='discover']/machine/world")
+    end
+
+    # Create a file inside the content directory under the stacks.local_document_cache_root
+    # @param [String] content The contents of the file to be created
+    # @param [String] filename The name of the file to be created
+    # @return [void]
+    def transfer_to_document_store(content, filename)
+      new_file = File.join(purl_druid.content_dir, filename)
+      File.open(new_file, 'w') { |f| f.write content }
+    end
+
+    def purl_druid
+      @purl_druid ||= DruidTools::PurlDruid.new item.pid, Config.stacks.local_document_cache_root
+    end
+
+    def prune_purl_dir
+      purl_druid.prune!
+    end
+
+    ##
+    # When publishing a PURL, we notify purl-fetcher of changes.
+    # If the purl service isn't configured, instead we drop a `aa11bb2222` file into the `local_recent_changes` folder
+    # to notify other applications watching the filesystem (i.e., purl-fetcher).
+    # We also remove any .deletes entry that may have left over from a previous removal
+    def publish_notify_on_success
+      id = item.pid.gsub(/^druid:/, '')
+      if Dor::Config.purl_services.url
+        purl_services = Dor::Config.purl_services.rest_client
+        purl_services["purls/#{id}"].post ''
+      else
+        Deprecation.warn(self, 'You have not configured perl-fetcher (Dor::Config.purl_services.url). This will result in an error in dor-services 7 ')
+        local_recent_changes = Config.stacks.local_recent_changes
+        raise ArgumentError, "Missing local_recent_changes directory: #{local_recent_changes}" unless File.directory?(local_recent_changes)
+
+        FileUtils.touch(File.join(local_recent_changes, id))
+        begin
+          DruidTools::Druid.new(id, Dor::Config.stacks.local_document_cache_root).deletes_delete_record
+        rescue Errno::EACCES
+          Dor.logger.warn "Access denied while trying to remove .deletes file for druid:#{id}"
+        end
+      end
+    end
+
+    ##
+    # When publishing a PURL, we notify purl-fetcher of changes.
+    def publish_delete_on_success
+      return unless Dor::Config.purl_services.url
+
+      id = item.pid.gsub(/^druid:/, '')
+
+      purl_services = Dor::Config.purl_services.rest_client
+      purl_services["purls/#{id}"].delete
+    end
+  end
+end