dor-services 5.2.0 → 5.3.0

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

@@ -8,7 +8,7 @@ handler = Class.new do
 
   def label(metadata)
     mods = Nokogiri::XML(metadata)
-    mods.root.add_namespace_definition('mods','http://www.loc.gov/mods/v3')
+    mods.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
   end
 

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

@@ -14,18 +14,18 @@ handler = Class.new do
     client = RestClient::Resource.new(Dor::Config.metadata.exist.url)
     response = client['db'].post(query, :content_type => 'application/xquery')
     doc = Nokogiri::XML(response)
-    doc.root.add_namespace_definition('exist','http://exist.sourceforge.net/NS/exist')
+    doc.root.add_namespace_definition('exist', 'http://exist.sourceforge.net/NS/exist')
     result = doc.xpath('/exist:result/*[1]').first
     result.nil? ? nil : result.to_s
   end
 
   def label(metadata)
     xml = Nokogiri::XML(metadata)
-    return "" if xml.root.nil?
+    return '' if xml.root.nil?
     case xml.root.name
     when 'msDesc' then xml.xpath('/msDesc/msIdentifier/collection').text
     when 'mods'   then
-      xml.root.add_namespace_definition('mods','http://www.loc.gov/mods/v3')
+      xml.root.add_namespace_definition('mods', 'http://www.loc.gov/mods/v3')
       xml.xpath('/mods:mods/mods:titleInfo[1]').xpath('mods:title|mods:nonSort').collect { |n| n.text }.join(' ').strip
     end
   end

data/lib/dor/services/metadata_service.rb

@@ -1,20 +1,19 @@
 require 'cache'
 
 module Dor
-
   class MetadataError < Exception ; end
 
-#  class MetadataHandler
-#
-#    def fetch(prefix, identifier)
-#      ### Return metadata for prefix/identifier combo
-#    end
-#
-#    def label(metadata)
-#      ### Return a Fedora-compatible label from the metadata format returned by #fetch
-#    end
-#
-#  end
+  #  class MetadataHandler
+  #
+  #    def fetch(prefix, identifier)
+  #      ### Return metadata for prefix/identifier combo
+  #    end
+  #
+  #    def label(metadata)
+  #      ### Return a Fedora-compatible label from the metadata format returned by #fetch
+  #    end
+  #
+  #  end
 
   class MetadataService
 
@@ -31,15 +30,15 @@ module Dor
         handler.prefixes.each do |prefix|
           handlers[prefix.to_sym] = handler
         end
-        return handler
+        handler
       end
 
       def known_prefixes
-        return handlers.keys
+        handlers.keys
       end
 
       def can_resolve?(identifier)
-        (prefix, identifier) = identifier.split(/:/,2)
+        (prefix, _identifier) = identifier.split(/:/, 2)
         handlers.keys.include?(prefix.to_sym)
       end
 
@@ -50,14 +49,14 @@ module Dor
 
       def fetch(identifier)
         @@cache.fetch(identifier) do
-          (prefix, identifier) = identifier.split(/:/,2)
+          (prefix, identifier) = identifier.split(/:/, 2)
           handler = handler_for(prefix)
           handler.fetch(prefix, identifier)
         end
       end
 
       def label_for(identifier)
-        (prefix, identifier) = identifier.split(/:/,2)
+        (prefix, identifier) = identifier.split(/:/, 2)
         handler = handler_for(prefix)
         handler.label(handler.fetch(prefix, identifier))
       end
@@ -65,10 +64,11 @@ module Dor
       def handler_for(prefix)
         handler = handlers[prefix.to_sym]
         raise MetadataError, "Unkown metadata prefix: #{prefix}" if handler.nil?
-        return handler
+        handler
       end
 
       private
+
       def handlers
         @handlers ||= {}
       end
@@ -76,9 +76,8 @@ module Dor
     end
 
   end
-
 end
 
-Dir[File.join(File.dirname(__FILE__),'metadata_handlers','*.rb')].each { |handler_file|
+Dir[File.join(File.dirname(__FILE__), 'metadata_handlers', '*.rb')].each { |handler_file|
   load handler_file
 }

data/lib/dor/services/registration_service.rb

@@ -4,75 +4,100 @@ module Dor
   class RegistrationService
 
     class << self
+
+      # @TODO: Why isn't all this logic in, for example, Dor::Item.create? or Dor::Base.create? or Dor::Creatable.create?
+      # @TODO: these duplicate checks could be combined into 1 query
+
+      # @param [String] pid an ID to check, if desired.  If not passed (or nil), a new ID is minted
+      # @return [String] a pid you can use immidately, either freshly minted or your checked value
+      # @raise [Dor::DuplicateIdError]
+      def unduplicated_pid(pid = nil)
+        return Dor::SuriService.mint_id unless pid
+        existing_pid = SearchService.query_by_id(pid).first
+        unless existing_pid.nil?
+          raise Dor::DuplicateIdError.new(existing_pid), "An object with the PID #{pid} has already been registered."
+        end
+        pid
+      end
+
+      # @param [String] source_id_string a fully qualified source:val or empty string
+      # @return [String] the same qualified source:id for immediate use
+      # @raise [Dor::DuplicateIdError]
+      def check_source_id(source_id_string)
+        return '' if source_id_string == ''
+        unless SearchService.query_by_id("#{source_id_string}").first.nil?
+          raise Dor::DuplicateIdError.new(source_id_string), "An object with the source ID '#{source_id_string}' has already been registered."
+        end
+        source_id_string
+      end
+
+      # @param [Hash{Symbol => various}] params
+      # @option params [String] :object_type required
+      # @option params [String] :label required
+      # @option params [String] :admin_policy required
+      # @option params [String] :metadata_source
+      # @option params [String] :rights
+      # @option params [String] :collection
+      # @option params [Hash{String => String}] :source_id Primary ID from another system, max one key/value pair!
+      # @option params [Hash] :other_ids including :uuid if known
+      # @option params [String] :pid Fully qualified PID if you don't want one generated for you
+      # @option params [Integer] :workflow_priority]
+      # @option params [Array<String>] :seed_datastream datastream_names
+      # @option params [Array<String>] :initiate_workflow workflow_ids
+      # @option params [Array] :tags
       def register_object(params = {})
         Dor.ensure_models_loaded!
         [:object_type, :label].each do |required_param|
-          raise Dor::ParameterError, "#{required_param.inspect} must be specified in call to #{self.name}.register_object" unless params[required_param]
+          raise Dor::ParameterError, "#{required_param.inspect} must be specified in call to #{name}.register_object" unless params[required_param]
         end
         metadata_source = params[:metadata_source]
-        if params[:label].length<1 && (metadata_source=='label' || metadata_source=='none')
-          raise Dor::ParameterError, "label cannot be empty to call #{self.name}.register_object"
+        if params[:label].length < 1 && (metadata_source == 'label' || metadata_source == 'none')
+          raise Dor::ParameterError, "label cannot be empty to call #{name}.register_object"
         end
         object_type = params[:object_type]
         item_class = Dor.registered_classes[object_type]
         raise Dor::ParameterError, "Unknown item type: '#{object_type}'" if item_class.nil?
 
-        content_model = params[:content_model]
-        admin_policy  = params[:admin_policy]
+        # content_model = params[:content_model]
+        # parent        = params[:parent]
         label         = params[:label]
         source_id     = params[:source_id] || {}
         other_ids     = params[:other_ids] || {}
         tags          = params[:tags] || []
-        parent        = params[:parent]
         collection    = params[:collection]
-        pid = nil
-        if params[:pid]
-          pid = params[:pid]
-          existing_pid = SearchService.query_by_id(pid).first
-          unless existing_pid.nil?
-            raise Dor::DuplicateIdError.new(existing_pid), "An object with the PID #{pid} has already been registered."
-          end
-        else
-          pid = Dor::SuriService.mint_id
-        end
 
-        rights=nil
+        # Check for sourceId conflict *before* potentially minting PID
+        source_id_string = check_source_id [source_id.keys.first, source_id[source_id.keys.first]].compact.join(':')
+        pid = unduplicated_pid(params[:pid])
+
+        raise ArgumentError, ":source_id Hash can contain at most 1 pair: recieved #{source_id.size}" if source_id.size > 1
+
+        rights = nil
         if params[:rights]
-          rights=params[:rights]
+          rights = params[:rights]
           unless %w(world stanford dark default none).include? rights
-            raise Dor::ParameterError, "Unknown rights setting '#{rights}' when calling #{self.name}.register_object"
-          end
-        end
-
-        source_id_string = [source_id.keys.first,source_id[source_id.keys.first]].compact.join(':')
-        unless source_id.empty?
-          existing_pid = SearchService.query_by_id("#{source_id_string}").first
-          unless existing_pid.nil?
-            raise Dor::DuplicateIdError.new(existing_pid), "An object with the source ID '#{source_id_string}' has already been registered."
+            raise Dor::ParameterError, "Unknown rights setting '#{rights}' when calling #{name}.register_object"
           end
         end
 
         if (other_ids.key?(:uuid) || other_ids.key?('uuid')) == false
           other_ids[:uuid] = UUIDTools::UUID.timestamp_create.to_s
         end
-        short_label = label.length>254 ? label[0,254] : label
-        apo_object = Dor.find(admin_policy, :lightweight => true)
-        adm_xml = apo_object.administrativeMetadata.ng_xml
-
+        apo_object = Dor.find(params[:admin_policy])
         new_item = item_class.new(:pid => pid)
-        new_item.label = short_label
+        new_item.label = label.length > 254 ? label[0, 254] : label
         idmd = new_item.identityMetadata
         idmd.sourceId = source_id_string
         idmd.add_value(:objectId, pid)
         idmd.add_value(:objectCreator, 'DOR')
         idmd.add_value(:objectLabel, label)
         idmd.add_value(:objectType, object_type)
-        other_ids.each_pair { |name,value| idmd.add_otherId("#{name}:#{value}") }
+        other_ids.each_pair { |name, value| idmd.add_otherId("#{name}:#{value}") }
         tags.each { |tag| idmd.add_value(:tag, tag) }
         new_item.admin_policy_object = apo_object
 
-        adm_xml.xpath('/administrativeMetadata/relationships/*').each do |rel|
-          short_predicate = ActiveFedora::RelsExtDatastream.short_predicate rel.namespace.href+rel.name
+        apo_object.administrativeMetadata.ng_xml.xpath('/administrativeMetadata/relationships/*').each do |rel|
+          short_predicate = ActiveFedora::RelsExtDatastream.short_predicate rel.namespace.href + rel.name
           if short_predicate.nil?
             ix = 0
             ix += 1 while ActiveFedora::Predicates.predicate_mappings[rel.namespace.href].key?(short_predicate = :"extra_predicate_#{ix}")
@@ -82,21 +107,21 @@ module Dor
         end
         new_item.add_collection(collection) if collection
         if rights && %w(item collection).include?(object_type)
-          rights_xml=apo_object.defaultObjectRights.ng_xml
-          new_item.datastreams['rightsMetadata'].content=rights_xml.to_s
+          rights_xml = apo_object.defaultObjectRights.ng_xml
+          new_item.datastreams['rightsMetadata'].content = rights_xml.to_s
           new_item.set_read_rights(rights) unless rights == 'default'    # already defaulted to default!
         end
-        #create basic mods from the label
-        if (metadata_source=='label')
-          ds=new_item.build_datastream('descMetadata');
+        # create basic mods from the label
+        if metadata_source == 'label'
+          ds = new_item.build_datastream('descMetadata')
           builder = Nokogiri::XML::Builder.new { |xml|
-            xml.mods( 'xmlns' => 'http://www.loc.gov/mods/v3', 'xmlns:xsi' => 'http://www.w3.org/2001/XMLSchema-instance',:version => '3.3', "xsi:schemaLocation" => 'http://www.loc.gov/mods/v3 http://www.loc.gov/standards/mods/v3/mods-3-3.xsd'){
-              xml.titleInfo{
+            xml.mods( 'xmlns' => 'http://www.loc.gov/mods/v3', 'xmlns:xsi' => 'http://www.w3.org/2001/XMLSchema-instance', :version => '3.3', 'xsi:schemaLocation' => 'http://www.loc.gov/mods/v3 http://www.loc.gov/standards/mods/v3/mods-3-3.xsd') {
+              xml.titleInfo {
                 xml.title label
               }
             }
           }
-          ds.content=builder.to_xml
+          ds.content = builder.to_xml
         end
 
         workflow_priority = params[:workflow_priority] ? params[:workflow_priority].to_i : 0
@@ -106,22 +131,18 @@ module Dor
 
         new_item.assert_content_model
 
-        new_item.class.ancestors.select { |x| x.respond_to? :to_class_uri }.each do |parent_class|
-          next if parent_class == ActiveFedora::Base
+        new_item.class.ancestors.select { |x| x.respond_to?(:to_class_uri) && x != ActiveFedora::Base }.each do |parent_class|
           new_item.add_relationship(:has_model, parent_class.to_class_uri)
         end
 
         new_item.save
-        begin
-          new_item.update_index if ::ENABLE_SOLR_UPDATES
-        rescue StandardError => e
-          Dor.logger.warn "Dor::RegistrationService.register_object failed to update solr index for #{new_item.pid}: #<#{e.class.name}: #{e.message}>"
-        end
-        return(new_item)
+        new_item
       end
 
+      # @param [Hash] params
+      # @see register_object similar but different
       def create_from_request(params)
-        other_ids = Array(params[:other_id]).collect do |id|
+        other_ids = Array(params[:other_id]).map do |id|
           if id =~ /^symphony:(.+)$/
             "#{$1.length < 14 ? 'catkey' : 'barcode'}:#{$1}"
           else
@@ -153,21 +174,19 @@ module Dor
           :collection         => params[:collection],
           :workflow_priority  => params[:workflow_priority]
         }
-        dor_params.delete_if { |k,v| v.nil? }
+        dor_params.delete_if { |k, v| v.nil? }
 
-        dor_obj = self.register_object(dor_params)
+        dor_obj = register_object(dor_params)
         pid = dor_obj.pid
-        location = URI.parse(Dor::Config.fedora.safeurl.sub(/\/*$/,'/')).merge("objects/#{pid}").to_s
-        reg_response = dor_params.dup.merge({ :location => location, :pid => pid })
+        location = URI.parse(Dor::Config.fedora.safeurl.sub(/\/*$/, '/')).merge("objects/#{pid}").to_s
+        dor_params.dup.merge({ :location => location, :pid => pid })
       end
 
       private
+
       def ids_to_hash(ids)
-        if ids.nil?
-          nil
-        else
-          Hash[Array(ids).collect { |id| id.split(/:/) }]
-        end
+        return nil if ids.nil?
+        Hash[Array(ids).map { |id| id.split(/:/) }]
       end
     end
   end

data/lib/dor/services/reset_workspace_service.rb

@@ -1,30 +1,26 @@
 module Dor
-
   # Rename the druid trees  at the end of the accessionWF in order to be cleaned/deleted later.
   class ResetWorkspaceService
 
     def self.reset_workspace_druid_tree(druid, version, workspace_root)
-
       druid_tree_path = DruidTools::Druid.new(druid, workspace_root).pathname.to_s
 
-      raise "The archived directory #{druid_tree_path}_v#{version} already existed." if  File.exists?("#{druid_tree_path}_v#{version}")
+      raise "The archived directory #{druid_tree_path}_v#{version} already existed." if  File.exist?("#{druid_tree_path}_v#{version}")
 
-      if File.exists?(druid_tree_path)
+      if File.exist?(druid_tree_path)
         FileUtils.mv(druid_tree_path, "#{druid_tree_path}_v#{version}")
-      end #Else is a truncated tree where we shouldn't do anything
-
+      end # Else is a truncated tree where we shouldn't do anything
     end
 
     def self.reset_export_bag(druid, version, export_root)
-
       id = druid.split(':').last
       bag_dir = File.join(export_root, id)
 
-      raise "The archived bag #{bag_dir}_v#{version} already existed." if  File.exists?("#{bag_dir}_v#{version}")
+      raise "The archived bag #{bag_dir}_v#{version} already existed." if  File.exist?("#{bag_dir}_v#{version}")
 
-      FileUtils.mv(bag_dir, "#{bag_dir}_v#{version}") if File.exists?(bag_dir)
+      FileUtils.mv(bag_dir, "#{bag_dir}_v#{version}") if File.exist?(bag_dir)
 
-      if File.exists?("#{bag_dir}.tar")
+      if File.exist?("#{bag_dir}.tar")
         FileUtils.mv("#{bag_dir}.tar", "#{bag_dir}_v#{version}.tar")
       end
     end

data/lib/dor/services/sdr_ingest_service.rb

@@ -6,9 +6,9 @@ module Dor
     # @param [Dor::Item] dor_item The representation of the digital object
     # @param [String] agreement_id  depreciated, included for backward compatability with common-accessoning
     # @return [void] Create the moab manifests, export data to a BagIt bag, kick off the SDR ingest workflow
-    def self.transfer(dor_item, agreement_id=nil)
+    def self.transfer(dor_item, agreement_id = nil)
       druid = dor_item.pid
-      workspace = DruidTools::Druid.new(druid,Dor::Config.sdr.local_workspace_root)
+      workspace = DruidTools::Druid.new(druid, Dor::Config.sdr.local_workspace_root)
       signature_catalog = get_signature_catalog(druid)
       new_version_id = signature_catalog.version_id + 1
       metadata_dir = extract_datastreams(dor_item, workspace)
@@ -20,14 +20,14 @@ module Dor
         content_dir = nil
       else
         new_file_list = content_addtions.path_list
-        content_dir = workspace.find_filelist_parent('content',new_file_list)
+        content_dir = workspace.find_filelist_parent('content', new_file_list)
       end
       content_group = version_inventory.group('content')
       unless content_group.nil? || content_group.files.empty?
         signature_catalog.normalize_group_signatures(content_group, content_dir)
       end
       # export the bag (in tar format)
-      bag_dir = Pathname(Dor::Config.sdr.local_export_home).join(druid.sub('druid:',''))
+      bag_dir = Pathname(Dor::Config.sdr.local_export_home).join(druid.sub('druid:', ''))
       bagger = Moab::Bagger.new(version_inventory, signature_catalog, bag_dir)
       bagger.reset_bag
       bagger.create_bag_inventory(:depositor)
@@ -38,7 +38,7 @@ module Dor
       # Now bootstrap SDR workflow. but do not create the workflows datastream
       dor_item.initialize_workflow('sdrIngestWF', false)
     rescue Exception => e
-      raise LyberCore::Exceptions::ItemError.new(druid, "Export failure", e)
+      raise LyberCore::Exceptions::ItemError.new(druid, 'Export failure', e)
     end
 
     # @param [String] druid The object identifier
@@ -57,11 +57,11 @@ module Dor
     # @return [Pathname] Pull all the datastreams specified in the configuration file
     #   into the workspace's metadata directory, overwriting existing file if present
     def self.extract_datastreams(dor_item, workspace)
-      metadata_dir = Pathname.new(workspace.path('metadata',create=true))
+      metadata_dir = Pathname.new(workspace.path('metadata', true))
       Config.sdr.datastreams.to_hash.each_pair do |ds_name, required|
         ds_name = ds_name.to_s
         metadata_file = metadata_dir.join("#{ds_name}.xml")
-        metadata_string = self.get_datastream_content(dor_item, ds_name, required)
+        metadata_string = get_datastream_content(dor_item, ds_name, required)
         metadata_file.open('w') { |f| f << metadata_string } if metadata_string
       end
       metadata_dir
@@ -76,7 +76,7 @@ module Dor
       ds = (ds_name == 'relationshipMetadata' ? 'RELS-EXT' : ds_name)
       if dor_item.datastreams.keys.include?(ds) && !dor_item.datastreams[ds].new?
         return dor_item.datastreams[ds].content
-      elsif (required == 'optional')
+      elsif required == 'optional'
         return nil
       else
         raise "required datastream #{ds_name} not found in DOR"
@@ -86,7 +86,7 @@ module Dor
     # @param [Pathname] metadata_dir the location of the metadata directory in the workspace
     # @param [Integer] expected the version identifer expected to be used in the versionMetadata
     def self.verify_version_metadata(metadata_dir, expected)
-      vmfile = metadata_dir.join("versionMetadata.xml")
+      vmfile = metadata_dir.join('versionMetadata.xml')
       verify_version_id(vmfile, expected, vmfile_version_id(vmfile))
       true
     end
@@ -95,7 +95,7 @@ module Dor
     # @param [Integer] expected The version number that should be in the file
     # @param [Integer] found The version number that is actually in the file
     def self.verify_version_id(pathname, expected, found)
-      raise "Version mismatch in #{pathname}, expected #{expected}, found #{found}" unless (expected == found)
+      raise "Version mismatch in #{pathname}, expected #{expected}, found #{found}" unless expected == found
       true
     end
 
@@ -104,7 +104,7 @@ module Dor
     def self.vmfile_version_id(pathname)
       verify_pathname(pathname)
       doc = Nokogiri::XML(File.open(pathname.to_s))
-      nodeset = doc.xpath("/versionMetadata/version")
+      nodeset = doc.xpath('/versionMetadata/version')
       version_id = nodeset.last['versionId']
       version_id.nil? ? nil : version_id.to_i
     end
@@ -127,9 +127,9 @@ module Dor
     def self.get_content_inventory(metadata_dir, druid, version_id)
       content_metadata = get_content_metadata(metadata_dir)
       if content_metadata
-        Stanford::ContentInventory.new.inventory_from_cm(content_metadata, druid, subset='preserve', version_id)
+        Stanford::ContentInventory.new.inventory_from_cm(content_metadata, druid, 'preserve', version_id)
       else
-        FileInventory.new(:type=>"version",:digital_object_id=>druid, :version_id=>version_id)
+        FileInventory.new(:type => 'version', :digital_object_id => druid, :version_id => version_id)
       end
     end
 
@@ -143,7 +143,7 @@ module Dor
     # @param [Pathname] metadata_dir The location of the the object's metadata files
     # @return [Moab::FileGroup] Traverse the metadata directory and generate a metadata group
     def self.get_metadata_file_group(metadata_dir)
-      file_group = FileGroup.new(:group_id=>'metadata').group_from_directory(metadata_dir)
+      file_group = FileGroup.new(:group_id => 'metadata').group_from_directory(metadata_dir)
       file_group
     end
 
@@ -158,7 +158,7 @@ module Dor
       verify_pathname(bag_dir.join('tagmanifest-sha256.txt'))
       verify_pathname(bag_dir.join('versionAdditions.xml'))
       verify_pathname(bag_dir.join('versionInventory.xml'))
-      verify_pathname(bag_dir.join('data','metadata','versionMetadata.xml'))
+      verify_pathname(bag_dir.join('data', 'metadata', 'versionMetadata.xml'))
       true
     end
 
@@ -170,5 +170,4 @@ module Dor
     end
 
   end
-
 end