dor-services 5.2.0 → 5.3.0

data/lib/dor/datastreams/datastream_spec_solrizer.rb

@@ -9,7 +9,7 @@ module DatastreamSpecSolrizer
       0
     end
     v = versionID.nil? ? '0' : versionID.to_s.split(/\./).last
-    [dsid,controlGroup,mimeType,v,s,label].join("|")
+    [dsid, controlGroup, mimeType, v, s, label].join('|')
   end
 end
 

data/lib/dor/datastreams/default_object_rights_ds.rb

@@ -6,45 +6,128 @@ module Dor
       t.copyright :path => 'copyright/human', :index_as => [:symbol]
       t.use_statement :path => '/use/human[@type=\'useAndReproduction\']', :index_as => [:symbol]
 
-      t.use do
-        t.machine
-        t.human
-      end
+      # t.use do
+      #   t.machine
+      #   t.human
+      # end
 
-      t.creative_commons :path => '/use/machine', :type => 'creativeCommons'
+      t.creative_commons :path => '/use/machine[@type=\'creativeCommons\']', :type => 'creativeCommons' do
+        t.uri :path => '@uri'
+      end
       t.creative_commons_human :path => '/use/human[@type=\'creativeCommons\']'
+      t.open_data_commons :path => '/use/machine[@type=\'openDataCommons\']', :type => 'openDataCommons' do
+        t.uri :path => '@uri'
+      end
+      t.open_data_commons_human :path => '/use/human[@type=\'openDataCommons\']'
     end
 
     define_template :creative_commons do |xml|
       xml.use {
-        xml.human(:type => "creativeCommons")
-        xml.machine(:type => "creativeCommons")
+        xml.human(:type => 'creativeCommons')
+        xml.machine(:type => 'creativeCommons', :uri => '')
+      }
+    end
+
+    define_template :open_data_commons do |xml|
+      xml.use {
+        xml.human(:type => 'openDataCommons')
+        xml.machine(:type => 'openDataCommons', :uri => '')
+      }
+    end
+
+    define_template :copyright do |xml|
+      xml.copyright {
+        xml.human
+      }
+    end
+
+    define_template :use_statement do |xml|
+      xml.use {
+        xml.human(type: 'useAndReproduction')
       }
     end
 
     def self.xml_template
       Nokogiri::XML::Builder.new do |xml|
-        xml.rightsMetadata{
-          xml.access(:type => 'discover'){
-            xml.machine{
+        xml.rightsMetadata {
+          xml.access(:type => 'discover') {
+            xml.machine {
               xml.world
             }
           }
-          xml.access(:type => 'read'){
-            xml.machine{
+          xml.access(:type => 'read') {
+            xml.machine {
               xml.world
             }
           }
-          xml.use{
+          xml.use {
             xml.human(:type => 'useAndReproduction')
-            xml.human(:type => "creativeCommons")
-            xml.machine(:type => "creativeCommons")
+            xml.human(:type => 'creativeCommons')
+            xml.machine(:type => 'creativeCommons', :uri => '')
+            xml.human(:type => 'openDataCommons')
+            xml.machine(:type => 'openDataCommons', :uri => '')
           }
-          xml.copyright{
+          xml.copyright {
             xml.human
           }
         }
       end.doc
     end
+
+    # Ensures that the template is present for the given term
+    def initialize_term!(term)
+      return unless find_by_terms(term).length < 1
+      ng_xml_will_change!
+      add_child_node(ng_xml.root, term)
+    end
+
+    # Assigns the defaultObjectRights object's term with the given value. Supports setting value to nil
+    def update_term!(term, val)
+      ng_xml_will_change!
+      if val.blank?
+        update_values({ [ term ] => nil })
+      else
+        initialize_term! term
+        update_values({ [ term ] => val })
+      end
+      normalize!
+    end
+
+    def content
+      ng_xml.human
+    end
+
+    # Purge the XML of any empty or duplicate elements -- keeps <rightsMetadata> clean
+    def normalize!
+      ng_xml_will_change!
+      doc = ng_xml
+      if doc.xpath('/rightsMetadata/use').length > 1
+        # <use> node needs consolidation
+        nodeset = doc.xpath('/rightsMetadata/use')
+        nodeset[1..-1].each do |node|
+          node.children.each do |child|
+            nodeset[0] << child # copy over to first <use> element
+          end
+          node.remove
+        end
+        fail unless doc.xpath('/rightsMetadata/use').length == 1
+      end
+
+      # Call out to the general purpose XML normalization service
+      ::Normalizer.new.tap do |norm|
+        norm.remove_empty_attributes(doc.root)
+        # cleanup ordering is important here
+        doc.xpath('//machine/text()').each { |node| node.content = node.content.strip }
+        doc.xpath('//human')
+          .tap { |nodeset| norm.clean_linefeeds(nodeset) }
+          .each do |node|
+            norm.trim_text(node)
+            norm.remove_empty_nodes(node)
+          end
+        doc.xpath('/rightsMetadata/copyright').each { |node| norm.remove_empty_nodes(node) }
+        doc.xpath('/rightsMetadata/use').each { |node| norm.remove_empty_nodes(node) }
+      end
+      self.content = doc.human
+    end
   end
 end

data/lib/dor/datastreams/desc_metadata_ds.rb

@@ -1,42 +1,45 @@
 module Dor
-class DescMetadataDS < ActiveFedora::OmDatastream
+  class DescMetadataDS < ActiveFedora::OmDatastream
 
-  MODS_NS = 'http://www.loc.gov/mods/v3'
-  set_terminology do |t|
-    t.root :path => 'mods', :xmlns => MODS_NS, :index_as => [:not_searchable]
-    t.originInfo  :index_as => [:not_searchable] do
-      t.publisher :index_as => [:stored_searchable]
-      t.date_created :path => 'dateCreated', :index_as => [:stored_searchable]
-      t.place :index_as => [:not_searchable] do
-        t.placeTerm :attributes => {:type => 'text'}, :index_as => [:stored_searchable]
+    MODS_NS = 'http://www.loc.gov/mods/v3'
+    set_terminology do |t|
+      t.root :path => 'mods', :xmlns => MODS_NS, :index_as => [:not_searchable]
+      t.originInfo  :index_as => [:not_searchable] do
+        t.publisher :index_as => [:stored_searchable]
+        t.date_created :path => 'dateCreated', :index_as => [:stored_searchable]
+        t.place :index_as => [:not_searchable] do
+          t.placeTerm :attributes => {:type => 'text'}, :index_as => [:stored_searchable]
+        end
       end
-    end
-    t.subject(:index_as => [:not_searchable]) do
-      t.geographic :index_as => [:symbol, :stored_searchable]
-      t.topic      :index_as => [:symbol, :stored_searchable]
-      t.temporal   :index_as => [:stored_searchable]
-    end
-    t.title_info(:path=>"titleInfo") {
-      t.main_title(:index_as=>[:symbol], :path=>"title", :label=>"title") {
-        t.main_title_lang(:path=>{:attribute=> "xml:lang"})
+      t.subject(:index_as => [:not_searchable]) do
+        t.geographic :index_as => [:symbol, :stored_searchable]
+        t.topic      :index_as => [:symbol, :stored_searchable]
+        t.temporal   :index_as => [:stored_searchable]
+      end
+      t.title_info(:path => 'titleInfo') {
+        t.main_title(:index_as => [:symbol], :path => 'title', :label => 'title') {
+          t.main_title_lang(:path => {:attribute => 'xml:lang'})
+        }
       }
-    }
-    t.coordinates :index_as => [:symbol]
-    t.extent      :index_as => [:symbol]
-    t.scale       :index_as => [:symbol]
-    t.topic       :index_as => [:symbol, :stored_searchable]
-    t.abstract    :index_as => [:stored_searchable]
-  end
+      t.language {
+        t.languageTerm :attributes => {:type => 'code', :authority => 'iso639-2b'}, :index_as => [:not_searchable]
+      }
+      t.coordinates :index_as => [:symbol]
+      t.extent      :index_as => [:symbol]
+      t.scale       :index_as => [:symbol]
+      t.topic       :index_as => [:symbol, :stored_searchable]
+      t.abstract    :index_as => [:stored_searchable]
+    end
 
-  def self.xml_template
-    Nokogiri::XML::Builder.new do |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.title
+    def self.xml_template
+      Nokogiri::XML::Builder.new do |xml|
+        xml.mods( 'xmlns' => MODS_NS, '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
+          }
         }
-      }
-    end.doc
-  end
+      end.doc
+    end
 
-end
+  end
 end

data/lib/dor/datastreams/embargo_metadata_ds.rb

@@ -4,13 +4,13 @@ class EmbargoMetadataDS < ActiveFedora::OmDatastream
   before_create :ensure_non_versionable
 
   set_terminology do |t|
-    t.root(:path => "embargoMetadata")
+    t.root(:path => 'embargoMetadata')
     t.status
     t.embargo_status(:path => 'status', :index_as => [:symbol])
-    t.release_date(:path => "releaseDate", :index_as => [:dateable])
-    t.release_access(:path => "releaseAccess")
-    t.twenty_pct_status( :path => "twentyPctVisibilityStatus", :index_as => [:symbol])
-    t.twenty_pct_release_date(:path => "twentyPctVisibilityReleaseDate")
+    t.release_date(:path => 'releaseDate', :index_as => [:dateable])
+    t.release_access(:path => 'releaseAccess')
+    t.twenty_pct_status(:path => 'twentyPctVisibilityStatus', :index_as => [:symbol])
+    t.twenty_pct_release_date(:path => 'twentyPctVisibilityReleaseDate')
   end
 
   # Default EmbargoMetadataDS xml
@@ -24,21 +24,21 @@ class EmbargoMetadataDS < ActiveFedora::OmDatastream
         xml.twentyPctVisibilityReleaseDate
       }
     end
-    return builder.doc
+    builder.doc
   end
 
-  def to_solr solr_doc = {}, *args
+  def to_solr(solr_doc = {}, *args)
     super
     #::Solrizer.insert_field(solr_doc, field_name, value, *index_types)
-    rd1  = self.release_date
-    rd20 = self.twenty_pct_release_date
+    rd1  = release_date
+    rd20 = twenty_pct_release_date
     ::Solrizer.insert_field(solr_doc, 'embargo_release', rd1.utc.strftime('%FT%TZ'), :dateable) unless rd1.nil?
     ::Solrizer.insert_field(solr_doc, 'twenty_pct_visibility_release', rd20.utc.strftime('%FT%TZ'), :dateable) unless rd20.nil?
     solr_doc
   end
 
   def ensure_non_versionable
-    self.versionable = "false"
+    self.versionable = 'false'
   end
 
   #################################################################################
@@ -56,16 +56,16 @@ class EmbargoMetadataDS < ActiveFedora::OmDatastream
 
   # Sets the release date.  Converts the date to beginning-of-day, UTC to help with Solr indexing
   # @param [Time] rd A Time object represeting the release date.  By default, it is set to now
-  def release_date=(rd=Time.now)
+  def release_date=(rd = Time.now.utc)
     update_values([:release_date] => rd.beginning_of_day.utc.xmlschema)
-    self.content=self.ng_xml.to_s
+    self.content = ng_xml.to_s
   end
 
   # Current releaseDate value
   # @return [Time]
   def release_date
     rd = term_values(:release_date).first
-    return (rd.nil? || rd.empty?) ? nil : Time.parse(rd)
+    (rd.nil? || rd.empty?) ? nil : Time.parse(rd)
   end
 
   def twenty_pct_status=(new_status)
@@ -79,7 +79,7 @@ class EmbargoMetadataDS < ActiveFedora::OmDatastream
 
   # Sets the 20% visibility release date.  Converts the date to beginning-of-day, UTC to help with Solr indexing
   # @param [Time] rd A Time object represeting the release date.  By default, it is set to now
-  def twenty_pct_release_date=(rd=Time.now)
+  def twenty_pct_release_date=(rd = Time.now.utc)
     update_values([:twenty_pct_release_date] => rd.beginning_of_day.utc.xmlschema)
     content_will_change!
   end
@@ -88,7 +88,7 @@ class EmbargoMetadataDS < ActiveFedora::OmDatastream
   # @return [Time]
   def twenty_pct_release_date
     rd = term_values(:twenty_pct_release_date).first
-    return (rd.nil? || rd.empty?) ? nil : Time.parse(rd)
+    (rd.nil? || rd.empty?) ? nil : Time.parse(rd)
   end
 
   # @return [Nokogiri::XML::Element] The releaseAccess node
@@ -100,7 +100,7 @@ class EmbargoMetadataDS < ActiveFedora::OmDatastream
   # @param [Nokogiri::XML::Document] new_doc Document that will replace the existing releaseAccess node
   def release_access_node=(new_doc)
     if (new_doc.root.name != 'releaseAccess')
-      raise "Trying to replace releaseAccess with a non-releaseAccess document"
+      raise 'Trying to replace releaseAccess with a non-releaseAccess document'
     end
 
     term_value_delete(:select => '//embargoMetadata/releaseAccess')

data/lib/dor/datastreams/events_ds.rb

@@ -17,7 +17,7 @@ class EventsDS < ActiveFedora::OmDatastream
     builder = Nokogiri::XML::Builder.new do |xml|
       xml.events
     end
-    return builder.doc
+    builder.doc
   end
 
   def ensure_non_versionable
@@ -30,7 +30,7 @@ class EventsDS < ActiveFedora::OmDatastream
   # @param [String] message what happened. Sets the content of the event with this message
   def add_event(type, who, message)
     ev = ng_xml.create_element 'event', message,
-      :type => type, :who => who, :when => Time.now.xmlschema
+      :type => type, :who => who, :when => Time.now.utc.xmlschema
     ng_xml.root.add_child(ev)
     ng_xml_will_change!
   end

data/lib/dor/datastreams/geo_metadata_ds.rb

@@ -20,7 +20,7 @@ module Dor
     }
 
     # hash with all namespaces
-    XMLNS = Hash[NS.map {|k,v| ["xmlns:#{k}", v]}]
+    XMLNS = Hash[NS.map {|k, v| ["xmlns:#{k}", v]}]
 
     # schema locations
     NS_XSD = NS.keys.collect {|k| "#{NS[k]} #{NS[k]}/#{k}.xsd"}
@@ -40,11 +40,8 @@ module Dor
     #     or nil if not provided
     def feature_catalogue
       root = ng_xml.xpath('/rdf:RDF/rdf:Description/gfc:FC_FeatureCatalogue', XMLNS)
-      if root.nil? || root.empty?
-        nil # Feature catalog is optional
-      else
-        Nokogiri::XML(root.first.to_xml)
-      end
+      return nil if root.nil? || root.empty?  # Feature catalog is optional
+      Nokogiri::XML(root.first.to_xml)
     end
 
     # @return [Nokogiri::XML::Document] Contains skeleton geoMetadata XML
@@ -54,7 +51,7 @@ module Dor
       Nokogiri::XML::Builder.new do |xml|
         xml['rdf'].RDF XMLNS,
           'xmlns:xsi' => 'http://www.w3.org/2001/XMLSchema-instance',
-          "xsi:schemaLocation" => NS_XSD.join(' ') do
+          'xsi:schemaLocation' => NS_XSD.join(' ') do
           xml['rdf'].Description 'rdf:about' => nil do
             xml['gmd'].MD_Metadata
           end
@@ -69,9 +66,7 @@ module Dor
     #      with .w, .e, .n., .s for west, east, north, south as floats
     def to_bbox
       params = { 'xmlns:gmd' => NS[:gmd], 'xmlns:gco' => NS[:gco] }
-      bb = metadata.xpath(
-        '//gmd:EX_Extent/gmd:geographicElement' +
-        '/gmd:EX_GeographicBoundingBox', params).first
+      bb = metadata.xpath('//gmd:EX_Extent/gmd:geographicElement/gmd:EX_GeographicBoundingBox', params).first
       Struct.new(:w, :e, :n, :s).new(
         bb.xpath('gmd:westBoundLongitude/gco:Decimal', params).text.to_f,
         bb.xpath('gmd:eastBoundLongitude/gco:Decimal', params).text.to_f,

data/lib/dor/datastreams/identity_metadata_ds.rb

@@ -3,7 +3,7 @@ class IdentityMetadataDS < ActiveFedora::OmDatastream
   include SolrDocHelper
 
   set_terminology do |t|
-    t.root(:path=>"identityMetadata")
+    t.root(:path => 'identityMetadata')
     t.objectId   :index_as => [:symbol]
     t.objectType :index_as => [:symbol]
     t.objectLabel
@@ -19,7 +19,7 @@ class IdentityMetadataDS < ActiveFedora::OmDatastream
     t.adminPolicy   :index_as => [:not_searchable]
   end
 
-  define_template :value do |builder,name,value,attrs|
+  define_template :value do |builder, name, value, attrs|
     builder.send(name.to_sym, value, attrs)
   end
 
@@ -27,21 +27,21 @@ class IdentityMetadataDS < ActiveFedora::OmDatastream
     Nokogiri::XML('<identityMetadata/>')
   end
 
-  def add_value(name, value, attrs={})
+  def add_value(name, value, attrs = {})
     add_child_node(ng_xml.root, :value, name, value, attrs)
   end
 
   def objectId
-    self.find_by_terms(:objectId).text
+    find_by_terms(:objectId).text
   end
 
   def sourceId
-    node = self.find_by_terms(:sourceId).first
-    node ? [node['source'],node.text].join(':') : nil
+    node = find_by_terms(:sourceId).first
+    node ? [node['source'], node.text].join(':') : nil
   end
 
   def sourceId=(value)
-    node = self.find_by_terms(:sourceId).first
+    node = find_by_terms(:sourceId).first
     unless value.present?   # so setting it to '' is the same as removal: worth documenting maybe?
       node.remove unless node.nil?
       return nil
@@ -55,13 +55,13 @@ class IdentityMetadataDS < ActiveFedora::OmDatastream
   end
 
   def tags
-    self.ng_xml.search('//tag').collect(&:content)
+    ng_xml.search('//tag').collect(&:content)
   end
 
   def otherId(type = nil)
-    result = self.find_by_terms(:otherId).to_a
+    result = find_by_terms(:otherId).to_a
     if type.nil?
-      result.collect { |n| [n['name'],n.text].join(':') }
+      result.collect { |n| [n['name'], n.text].join(':') }
     else
       result.select { |n| n['name'] == type }.collect { |n| n.text }
     end
@@ -75,7 +75,7 @@ class IdentityMetadataDS < ActiveFedora::OmDatastream
     node
   end
 
-  def to_solr(solr_doc=Hash.new, *args)
+  def to_solr(solr_doc = {}, *args)
     super(solr_doc, *args)
 
     if digital_object.respond_to?(:profile)
@@ -86,22 +86,22 @@ class IdentityMetadataDS < ActiveFedora::OmDatastream
 
     if sourceId.present?
       (name, id) = sourceId.split(/:/, 2)
-      add_solr_value(solr_doc, "dor_id", id, :symbol, [:stored_searchable])
-      add_solr_value(solr_doc, "identifier", sourceId, :symbol, [:stored_searchable])
-      add_solr_value(solr_doc, "source_id", sourceId, :symbol, [])
+      add_solr_value(solr_doc, 'dor_id', id, :symbol, [:stored_searchable])
+      add_solr_value(solr_doc, 'identifier', sourceId, :symbol, [:stored_searchable])
+      add_solr_value(solr_doc, 'source_id', sourceId, :symbol, [])
     end
     otherId.compact.each { |qid|
       # this section will solrize barcode and catkey, which live in otherId
       (name, id) = qid.split(/:/, 2)
-      add_solr_value(solr_doc, "dor_id", id, :symbol, [:stored_searchable])
-      add_solr_value(solr_doc, "identifier", qid, :symbol, [:stored_searchable])
+      add_solr_value(solr_doc, 'dor_id', id, :symbol, [:stored_searchable])
+      add_solr_value(solr_doc, 'identifier', qid, :symbol, [:stored_searchable])
       add_solr_value(solr_doc, "#{name}_id", id, :symbol, [])
     }
 
     # do some stuff to make tags in general and project tags specifically more easily searchable and facetable
-    self.find_by_terms(:tag).each { |tag|
+    find_by_terms(:tag).each { |tag|
       (prefix, rest) = tag.text.split(/:/, 2)
-      prefix = prefix.downcase.strip.gsub(/\s/,'_')
+      prefix = prefix.downcase.strip.gsub(/\s/, '_')
       unless rest.nil?
         # this part will index a value in a field specific to the tag, e.g. registered_by_tag_*,
         # book_tag_*, project_tag_*, remediated_by_tag_*, etc.  project_tag_* and registered_by_tag_*
@@ -116,13 +116,13 @@ class IdentityMetadataDS < ActiveFedora::OmDatastream
       tag_parts = tag.text.split(/:/)
       progressive_tag_prefix = ''
       tag_parts.each_with_index do |part, index|
-        progressive_tag_prefix += " : " if index > 0
+        progressive_tag_prefix += ' : ' if index > 0
         progressive_tag_prefix += part.strip
-        add_solr_value(solr_doc, "exploded_tag", progressive_tag_prefix, :symbol, [])
+        add_solr_value(solr_doc, 'exploded_tag', progressive_tag_prefix, :symbol, [])
       end
     }
 
-    return solr_doc
+    solr_doc
   end
-end #class
+end # class
 end