puree 1.9.1 → 2.0.0

data/lib/puree/xml_extractor/mixins/owner_mixin.rb

@@ -0,0 +1,18 @@
+module Puree
+
+  module XMLExtractor
+
+    # Owner mixin.
+    #
+    module OwnerMixin
+
+      # @return [Puree::Model::OrganisationalUnitHeader, nil]
+      def owner
+        xpath_result = xpath_query '/managingOrganisationalUnit'
+        Puree::XMLExtractor::Shared.organisation_header xpath_result
+      end
+
+    end
+
+  end
+end

data/lib/puree/xml_extractor/mixins/peer_reviewed_mixin.rb

@@ -8,7 +8,7 @@ module Puree
 
       # @return [Boolean, nil]
       def peer_reviewed
-        xpath_result = xpath_query_for_single_value('/peerReview/peerReviewed')
+        xpath_result = xpath_query_for_single_value('/peerReview')
         return true if xpath_result === 'true'
         return false if xpath_result === 'false'
         nil

data/lib/puree/xml_extractor/mixins/person_mixin.rb

@@ -0,0 +1,45 @@
+module Puree
+
+  module XMLExtractor
+
+    # Person mixin.
+    #
+    module PersonMixin
+
+      private
+
+      # @return [Array<Puree::Model::EndeavourPerson>]
+      def persons(type, xpath_query_path)
+        xpath_result = xpath_query xpath_query_path
+        arr = []
+        xpath_result.each do |i|
+          uuid_internal = i.at_xpath('person/@uuid')
+          uuid_external = i.at_xpath('externalPerson/@uuid')
+          if uuid_internal
+            person_type = 'internal'
+            uuid = uuid_internal.text.strip
+          elsif uuid_external
+            person_type = 'external'
+            uuid = uuid_external.text.strip
+          else
+            person_type = 'other'
+            uuid = ''
+          end
+          if person_type === type
+            person = Puree::Model::EndeavourPerson.new
+            person.uuid = uuid
+            name = Puree::Model::PersonName.new
+            name.first = i.xpath('name/firstName').text.strip
+            name.last = i.xpath('name/lastName').text.strip
+            person.name = name if name.data?
+            person.role = i.xpath('personRole').text.strip
+            arr << person if person.data?
+          end
+        end
+        arr
+      end
+
+    end
+
+  end
+end

data/lib/puree/xml_extractor/mixins/publisher_mixin.rb

@@ -0,0 +1,22 @@
+module Puree
+
+  module XMLExtractor
+
+    # Publisher extractor mixin.
+    #
+    module PublisherMixin
+
+      # @return [Puree::Model::PublisherHeader, nil]
+      def publisher
+        xpath_result = xpath_query '/publisher'
+        h = Puree::Model::PublisherHeader.new
+        h.uuid = xpath_result.xpath('@uuid').text.strip
+        h.name = xpath_result.xpath('name').text.strip
+        h.type = xpath_result.xpath('type').text.strip
+        h.data? ? h : nil
+      end
+
+    end
+
+  end
+end

data/lib/puree/xml_extractor/mixins/{associated_mixin.rb → research_output_mixin.rb}

@@ -2,19 +2,19 @@ module Puree
 
   module XMLExtractor
 
-    # Associated extractor mixin.
+    # Research output extractor mixin.
     #
-    module AssociatedMixin
+    module ResearchOutputMixin
 
-      # Combines projects and publications
+      # Related research outputs
       # @return [Array<Puree::Model::RelatedContentHeader>]
-      def associated
-        xpath_result = xpath_query '/associatedContent/relatedContent'
+      def research_outputs
+        xpath_result = xpath_query '/relatedResearchOutputs/relatedResearchOutput'
         data_arr = []
         xpath_result.each { |i|
           related = Puree::Model::RelatedContentHeader.new
-          related.type = i.xpath('typeClassification').text.strip
-          related.title = i.xpath('title').text.strip
+          related.type = i.xpath('type').text.strip
+          related.title = i.xpath('name').text.strip
           related.uuid = i.attr('uuid').strip
           data_arr << related
         }

data/lib/puree/xml_extractor/mixins/title_mixin.rb

@@ -0,0 +1,17 @@
+module Puree
+
+  module XMLExtractor
+
+    # Title mixin.
+    #
+    module TitleMixin
+
+      # @return [String, nil]
+      def title
+        xpath_query_for_single_value '/title'
+      end
+
+    end
+
+  end
+end

data/lib/puree/xml_extractor/mixins/type_mixin.rb

@@ -0,0 +1,17 @@
+module Puree
+
+  module XMLExtractor
+
+    # Type mixin.
+    #
+    module TypeMixin
+
+      # @return [String, nil]
+      def type
+        xpath_query_for_single_value '/type'
+      end
+
+    end
+
+  end
+end

data/lib/puree/xml_extractor/mixins/workflow_mixin.rb

@@ -0,0 +1,17 @@
+module Puree
+
+  module XMLExtractor
+
+    # Workflow extractor mixin.
+    #
+    module WorkflowMixin
+
+      # @return [String, nil]
+      def workflow
+        xpath_query_for_single_value '/workflow'
+      end
+
+    end
+
+  end
+end

data/lib/puree/xml_extractor/organisational_unit.rb

@@ -0,0 +1,82 @@
+module Puree
+
+  module XMLExtractor
+
+    # Organisational unit XML extractor.
+    #
+    class OrganisationalUnit < Puree::XMLExtractor::Resource
+      include Puree::XMLExtractor::TypeMixin
+
+      def initialize(xml)
+        super
+        setup_model :organisational_unit
+      end
+
+      # @return [Puree::Model::Address]
+      def address
+        xpath_result = xpath_query '/addresses/address'
+        if !xpath_result.empty?
+          a = Puree::Model::Address.new
+          street = xpath_result.xpath('street').text.strip
+          a.street = street unless street.empty?
+          building = xpath_result.xpath('building').text.strip
+          a.building = building unless building.empty?
+          postcode = xpath_result.xpath('postalCode').text.strip
+          a.postcode = postcode unless building.empty?
+          city = xpath_result.xpath('city').text.strip
+          a.city = city unless city.empty?
+          country = xpath_result.xpath('country').text.strip
+          a.country = country unless country.empty?
+          a
+        end
+      end
+
+      # @return [Array<String>]
+      def email_addresses
+        xpath_query_for_multi_value '/emails/email'
+      end
+
+      # @return [String, nil]
+      def name
+        xpath_query_for_single_value '/name'
+      end
+
+      # @return [Puree::Model::OrganisationalUnitHeader, nil]
+      def parent
+        xpath_result = xpath_query '/parents/parent'
+        Puree::XMLExtractor::Shared.organisation_header xpath_result
+      end
+
+      # @return [Array<String>]
+      def phone_numbers
+        xpath_query_for_multi_value '/phoneNumbers/phoneNumber'
+      end
+
+      # @return [Array<String>]
+      def urls
+        xpath_query_for_multi_value '/webAddresses/webAddress'
+      end
+
+      private
+
+      def xpath_root
+        '/organisationalUnit'
+      end
+
+      def combine_metadata
+        super
+        @model.address = address
+        @model.email_addresses = email_addresses
+        @model.name = name
+        @model.parent = parent
+        @model.phone_numbers = phone_numbers
+        @model.type = type
+        @model.urls = urls
+        @model
+      end      
+
+    end
+
+  end
+
+end

data/lib/puree/xml_extractor/paper.rb

@@ -1,14 +1,28 @@
 module Puree
   module XMLExtractor
 
-    # Paper XML extractor.
+    # Paper base XML extractor.
     #
-    class Paper < Puree::XMLExtractor::PaperBase
+    class Paper < Puree::XMLExtractor::ResearchOutput
+      include Puree::XMLExtractor::PagesMixin
+      include Puree::XMLExtractor::PageRangeMixin
+      include Puree::XMLExtractor::PeerReviewedMixin
 
-      def initialize(xml:)
+      def initialize(xml)
         super
+        setup_model :paper
       end
 
+      private
+
+      def combine_metadata
+        super
+        @model.pages = pages
+        @model.page_range = page_range
+        @model.peer_reviewed = peer_reviewed
+        @model
+      end      
+
     end
   end
 end

data/lib/puree/xml_extractor/person.rb

@@ -5,47 +5,44 @@ module Puree
     # Person XML extractor.
     #
     class Person < Puree::XMLExtractor::Resource
+      include Puree::XMLExtractor::IdentifierMixin
+      include Puree::XMLExtractor::KeywordMixin
 
-      def initialize(xml:)
+      def initialize(xml)
         super
-        @resource_type = :person
+        setup_model :person
       end
 
-      # @return [Array<Puree::Model::OrganisationHeader>]
+      # @return [Array<Puree::Model::OrganisationalUnitHeader>]
       def affiliations
-        xpath_result = xpath_query '//organisation'
-        Puree::XMLExtractor::Shared.organisation_multi_header xpath_result
+        xpath_result = xpath_query '/staffOrganisationAssociations/staffOrganisationAssociation/organisationalUnit'
+        Puree::XMLExtractor::Shared.organisation_multi_header xpath_result if xpath_result
       end
 
       # @return [Array<String>]
       def email_addresses
-        xpath_query_for_multi_value '//emails/classificationDefinedStringFieldExtension/value'
-      end
-
-      # @return [String, nil]
-      def employee_id
-        id '/dk/atira/pure/person/personsources/employee'
-      end
-
-      # @return [String, nil]
-      def hesa_id
-        id '/dk/atira/pure/person/personsources/hesastaff'
+        xpath_query_for_multi_value '/staffOrganisationAssociations/staffOrganisationAssociation/emails/email'
       end
 
       # @return [Array<String>]
       def image_urls
-        xpath_query_for_multi_value '/photos/file/url'
+        xpath_result = xpath_query '/profilePhotos/profilePhoto'
+        arr = []
+        xpath_result.each do |i|
+          arr << i.attr('url').strip
+        end
+        arr.uniq
       end
 
       # @return [Array<String>]
       def keywords
-        xpath_query_for_multi_value '//keywordGroup/keyword/userDefinedKeyword/freeKeyword'
+        keyword_group 'userDefinedKeywordContainers'
       end
 
       # @return [Puree::Model::PersonName, nil]
       def name
         xpath_result = xpath_query '/name'
-        if xpath_result
+        if !xpath_result.empty?
           first = xpath_result.xpath('firstName').text.strip
           last = xpath_result.xpath('lastName').text.strip
           model = Puree::Model::PersonName.new
@@ -60,26 +57,24 @@ module Puree
         xpath_query_for_single_value '/orcid'
       end
 
-      # @return [String, nil]
-      def scopus_id
-        id '/dk/atira/pure/person/personsources/scopusauthor'
-      end
-
       private
 
-      # @return [String, nil]
-      def id(uri)
-        xpath_result = xpath_query '/sources/classificationDefinedStringFieldExtension'
-        if xpath_result
-          xpath_result.each do |i|
-            if i.xpath('classification/uri').text.strip === uri
-              return i.xpath('value').text.strip
-            end
-          end
-        end
-        nil
+      def xpath_root
+        '/person'
       end
 
+      def combine_metadata
+        super
+        @model.affiliations = affiliations
+        @model.email_addresses = email_addresses
+        @model.identifiers = identifiers
+        @model.image_urls = image_urls
+        @model.keywords = keywords
+        @model.name = name
+        @model.orcid = orcid
+        @model
+      end      
+
     end
 
   end

data/lib/puree/xml_extractor/project.rb

@@ -5,12 +5,16 @@ module Puree
     # Project XML extractor.
     #
     class Project < Puree::XMLExtractor::Resource
-      # include Puree::XMLExtractor::AssociatedMixin # not present in stable API
-      include Puree::XMLExtractor::ExternalOrganisationsMixin
-
-      def initialize(xml:)
+      include Puree::XMLExtractor::ExternalOrganisationMixin
+      include Puree::XMLExtractor::IdentifierMixin
+      include Puree::XMLExtractor::OrganisationalUnitMixin
+      include Puree::XMLExtractor::PersonMixin
+      include Puree::XMLExtractor::TitleMixin
+      include Puree::XMLExtractor::TypeMixin
+
+      def initialize(xml)
         super
-        @resource_type = :project
+        setup_model :project
       end
 
       # @return [String, nil]
@@ -20,23 +24,10 @@ module Puree
 
       # @return [String, nil]
       def description
-        xpath_query_for_single_value '/description/localizedString'
-      end
-
-      # @return [Boolean]
-      def funded?
-        xpath_result = xpath_query_for_single_value '/type'
-        return false if xpath_result.downcase.include? 'nonfunded'
-        true
-      end
-
-      # @return [Array<Puree::Model::OrganisationHeader>]
-      def organisations
-        xpath_result = xpath_query '/organisations/association/organisation'
-        Puree::XMLExtractor::Shared.organisation_multi_header xpath_result
+        xpath_query_for_single_value '/descriptions/description'
       end
 
-      # @return [Puree::Model::OrganisationHeader, nil]
+      # @return [Puree::Model::OrganisationalUnitHeader, nil]
       def owner
         xpath_result = xpath_query '/owner'
         Puree::XMLExtractor::Shared.organisation_header xpath_result
@@ -44,86 +35,59 @@ module Puree
 
       # @return [Array<Puree::Model::EndeavourPerson>]
       def persons_internal
-        persons 'internal'
+        persons 'internal', '/participants/participant'
       end
 
       # @return [Array<Puree::Model::EndeavourPerson>]
       def persons_external
-        persons 'external'
+        persons 'external', '/participants/participant'
       end
 
       # @return [Array<Puree::Model::EndeavourPerson>]
       def persons_other
-        persons 'other'
+        persons 'other', '/participants/participant'
       end
 
       # @return [String, nil]
       def status
-        xpath_query_for_single_value '/status/term/localizedString'
+        xpath_query_for_single_value '/status'
       end
 
       # @return [Puree::Model::TemporalRange, nil]
-      def temporal_expected
-        temporal_range '/expectedStartDate', '/expectedEndDate'
-      end
-
-      # @return [Puree::Model::TemporalRange, nil]
-      def temporal_actual
-        temporal_range '/startFinishDate/startDate', '/startFinishDate/endDate'
-      end
-
-      # @return [String, nil]
-      def title
-        xpath_query_for_single_value '/title/localizedString'
-      end
-
-      # @return [String, nil]
-      def type
-        xpath_query_for_single_value '/typeClassification/term/localizedString'
+      def temporal
+        temporal_range '/period/startDate', '/period/endDate'
       end
 
       # @return [String, nil]
       def url
-        xpath_query_for_single_value '/projectURL'
+        xpath_query_for_single_value '/links/link/url'
       end
 
       private
 
-      # @return [Array<Endeavour::Person>]
-      def persons(type)
-        xpath_result = xpath_query '/persons/participantAssociation'
-        arr = []
-        xpath_result.each do |i|
-          uuid_internal = i.at_xpath('person/@uuid')
-          uuid_external = i.at_xpath('externalPerson/@uuid')
-          if uuid_internal
-            person_type = 'internal'
-            uuid = uuid_internal.text.strip
-          elsif uuid_external
-            person_type = 'external'
-            uuid = uuid_external.text.strip
-          else
-            person_type = 'other'
-            uuid = ''
-          end
-          if person_type === type
-            person = Puree::Model::EndeavourPerson.new
-            person.uuid = uuid
-
-            name = Puree::Model::PersonName.new
-            name.first = i.xpath('person/name/firstName').text.strip
-            name.last = i.xpath('person/name/lastName').text.strip
-            person.name = name
-
-            role = i.xpath('personRole/term/localizedString').text.strip
-            person.role = role
-
-            arr << person if person.data?
-          end
-        end
-        arr.uniq { |d| d.uuid }
+      def xpath_root
+        '/project'
       end
 
+      def combine_metadata
+        super
+        @model.acronym = acronym
+        @model.description = description
+        @model.external_organisations = external_organisations
+        @model.identifiers = identifiers
+        @model.organisational_units = organisational_units
+        @model.owner = owner
+        @model.persons_internal = persons_internal
+        @model.persons_external = persons_external
+        @model.persons_other = persons_other
+        @model.status = status
+        @model.temporal = temporal
+        @model.title = title
+        @model.type = type
+        @model.url = url
+        @model
+      end      
+
       # @return [Puree::Model::TemporalRange, nil]
       def temporal_range(start_path, end_path)
         range_start = xpath_query_for_single_value start_path

data/lib/puree/xml_extractor/publisher.rb

@@ -5,10 +5,11 @@ module Puree
     # Publisher XML extractor.
     #
     class Publisher < Puree::XMLExtractor::Resource
+      include Puree::XMLExtractor::TypeMixin
 
-      def initialize(xml:)
+      def initialize(xml)
         super
-        @resource_type = :publisher
+        setup_model :publisher
       end
 
       # @return [String, nil]
@@ -16,10 +17,18 @@ module Puree
         xpath_query_for_single_value '/name'
       end
 
-      # Adds no value as value is Publisher
-      # def type
-      #   xpath_query_for_single_value '/typeClassification/term/localizedString'
-      # end
+      private
+
+      def xpath_root
+        '/publisher'
+      end
+
+      def combine_metadata
+        super
+        @model.name = name
+        @model.type = type
+        @model
+      end
 
     end