puree 0.20.0 → 1.0.0

data/lib/puree/xml_extractor/download.rb

@@ -0,0 +1,42 @@
+module Puree
+
+  module XMLExtractor
+
+    # Download XML extractor.
+    #
+    class Download < Puree::XMLExtractor::Base
+
+      def initialize(xml:)
+        @resource_type = :download
+        super
+      end
+
+      # Statistic
+      #
+      # @return [Array<Puree::Model::DownloadHeader>]
+      def statistics
+        path = "#{service_response_name}/downloadCount"
+        xpath_result = @doc.xpath(path)
+        data_arr = []
+        xpath_result.each { |i|
+          model = Puree::Model::DownloadHeader.new
+          model.uuid = i.attr('uuid').strip
+          model.count = i.attr('downloads').strip.to_i
+          data_arr << model
+        }
+        data_arr.uniq { |d| d.uuid }
+      end
+
+      # Is there any data after get?
+      #
+      # @return [Boolean]
+      def get_data?
+        # TO DO Inefficient!
+        statistics.size ? true : false
+      end
+
+    end
+
+  end
+
+end

data/lib/puree/xml_extractor/event.rb

@@ -0,0 +1,63 @@
+module Puree
+
+  module XMLExtractor
+
+    # Event XML extractor.
+    #
+    class Event < Puree::XMLExtractor::Resource
+
+      def initialize(xml:)
+        super
+        @resource_type = :event
+      end
+
+      # @return [String, nil]
+      def city
+        xpath_query_for_single_value '/city'
+      end
+
+      # @return [String, nil]
+      def country
+        xpath_query_for_single_value '/country/term/localizedString'
+      end
+
+      # @return [Puree::Model::TemporalRange, nil]
+      def date
+        xpath_result = xpath_query '/dateRange'
+        range_start_str = xpath_result.xpath('startDate').text.strip
+        range_end_str = xpath_result.xpath('endDate').text.strip
+        if !range_start_str.empty?
+          range = Puree::Model::TemporalRange.new
+          range.start = Time.parse range_start_str
+          if !range_end_str.empty?
+            range.end = Time.parse range_end_str
+          end
+          range
+        end
+      end
+
+      # @return [String, nil]
+      def description
+        xpath_query_for_single_value '/description'
+      end
+
+      # @return [String, nil]
+      def location
+        xpath_query_for_single_value '/location'
+      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'
+      end
+
+    end
+
+  end
+
+end

data/lib/puree/xml_extractor/journal.rb

@@ -0,0 +1,33 @@
+module Puree
+
+  module XMLExtractor
+
+    # Journal XML extractor.
+    #
+    class Journal < Puree::XMLExtractor::Resource
+
+      def initialize(xml:)
+        super
+        @resource_type = :journal
+      end
+
+      # @return [String, nil]
+      def issn
+        xpath_query_for_single_value '/issns/issn/string'
+      end
+
+      # @return [String, nil]
+      def publisher
+        xpath_query_for_single_value '/publisher/name'
+      end
+
+      # @return [String, nil]
+      def title
+        xpath_query_for_single_value '/titles/title/string'
+      end
+
+    end
+
+  end
+
+end

data/lib/puree/xml_extractor/organisation.rb

@@ -0,0 +1,75 @@
+module Puree
+
+  module XMLExtractor
+
+    # Organisation XML extractor.
+    #
+    class Organisation < Puree::XMLExtractor::Resource
+
+      def initialize(xml:)
+        super
+        @resource_type = :organisation
+      end
+
+      # @return [Array<Puree::Model::Address>]
+      def address
+        xpath_result = xpath_query '/addresses/classifiedAddress'
+        data = []
+        xpath_result.each do |d|
+          a = Puree::Model::Address.new
+          street = d.xpath('street').text.strip
+          a.street = street unless street.empty?
+          building = d.xpath('building').text.strip
+          a.building = building unless building.empty?
+          postcode = d.xpath('postalCode').text.strip
+          a.postcode = postcode unless building.empty?
+          city = d.xpath('city').text.strip
+          a.city = city unless city.empty?
+          country = d.xpath('country/term/localizedString').text.strip
+          a.country = country unless country.empty?
+          data << a
+        end
+        data.uniq { |d| d.street }
+      end
+
+      # @return [Array<String>]
+      def email_addresses
+        xpath_query_for_multi_value '/emails/classificationDefinedStringFieldExtension/value'
+      end
+
+      # @return [String, nil]
+      def name
+        xpath_query_for_single_value '/name/localizedString'
+      end
+
+      # @return [Array<Puree::Model::OrganisationHeader>]
+      def organisations
+        xpath_result = xpath_query '/organisations/organisation'
+        Puree::XMLExtractor::Shared.organisation_multi_header xpath_result
+      end
+
+      # @return [Puree::Model::OrganisationHeader, nil]
+      def parent
+        organisations.first unless organisations.empty?
+      end
+
+      # @return [Array<String>]
+      def phone_numbers
+        xpath_query_for_multi_value '/phoneNumbers/classificationDefinedStringFieldExtension/value'
+      end
+
+      # @return [String, nil]
+      def type
+        xpath_query_for_single_value '/typeClassification/term/localizedString'
+      end
+
+      # @return [Array<String>]
+      def urls
+        xpath_query_for_multi_value '/webAddresses/classificationDefinedFieldExtension/value/localizedString'
+      end
+
+    end
+
+  end
+
+end

data/lib/puree/xml_extractor/person.rb

@@ -0,0 +1,57 @@
+module Puree
+
+  module XMLExtractor
+
+    # Person XML extractor.
+    #
+    class Person < Puree::XMLExtractor::Resource
+
+      def initialize(xml:)
+        super
+        @resource_type = :person
+      end
+
+      # @return [Array<Puree::Model::OrganisationHeader>]
+      def affiliations
+        xpath_result = xpath_query '//organisation'
+        Puree::XMLExtractor::Shared.organisation_multi_header xpath_result
+      end
+
+      # @return [Array<String>]
+      def email_addresses
+        xpath_query_for_multi_value '//emails/classificationDefinedStringFieldExtension/value'
+      end
+
+      # @return [Array<String>]
+      def image_urls
+        xpath_query_for_multi_value '/photos/file/url'
+      end
+
+      # @return [Array<String>]
+      def keywords
+        xpath_query_for_multi_value '//keywordGroup/keyword/userDefinedKeyword/freeKeyword'
+      end
+
+      # @return [Puree::Model::PersonName, nil]
+      def name
+        xpath_result = xpath_query '/name'
+        if xpath_result
+          first = xpath_result.xpath('firstName').text.strip
+          last = xpath_result.xpath('lastName').text.strip
+          model = Puree::Model::PersonName.new
+          model.first = first unless first.empty?
+          model.last = last unless last.empty?
+          model
+        end
+      end
+
+      # @return [String, nil]
+      def orcid
+        xpath_query_for_single_value '/orcid'
+      end
+
+    end
+
+  end
+
+end

data/lib/puree/xml_extractor/project.rb

@@ -0,0 +1,135 @@
+module Puree
+
+  module XMLExtractor
+
+    # Project XML extractor.
+    #
+    class Project < Puree::XMLExtractor::Resource
+
+      def initialize(xml:)
+        super
+        @resource_type = :project
+      end
+
+      # @return [String, nil]
+      def acronym
+        xpath_query_for_single_value '/acronym'
+      end
+
+      # @return [String, nil]
+      def description
+        xpath_query_for_single_value '/description/localizedString'
+      end
+
+      # @return [Array<Puree::Model::OrganisationHeader>]
+      def organisations
+        xpath_result = xpath_query '/organisations/association/organisation'
+        Puree::XMLExtractor::Shared.organisation_multi_header xpath_result
+      end
+
+      # @return [Puree::Model::OrganisationHeader, nil]
+      def owner
+        xpath_result = xpath_query '/owner'
+        Puree::XMLExtractor::Shared.organisation_header xpath_result
+      end
+
+      # @return [Array<Puree::Model::EndeavourPerson>]
+      def persons_internal
+        persons 'internal'
+      end
+
+      # @return [Array<Puree::Model::EndeavourPerson>]
+      def persons_external
+        persons 'external'
+      end
+
+      # @return [Array<Puree::Model::EndeavourPerson>]
+      def persons_other
+        persons 'other'
+      end
+
+      # @return [String, nil]
+      def status
+        xpath_query_for_single_value '/status/term/localizedString'
+      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'
+      end
+
+      # @return [String, nil]
+      def url
+        xpath_query_for_single_value '/projectURL'
+      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 }
+      end
+
+      # @return [Puree::Model::TemporalRange, nil]
+      def temporal_range(start_path, end_path)
+        range_start = xpath_query_for_single_value start_path
+        range_end = xpath_query_for_single_value end_path
+        if range_start || range_end
+          range = Puree::Model::TemporalRange.new
+          range.start = Time.new range_start if range_start
+          range.end = Time.new range_end if range_end
+          return range
+        end
+        nil
+      end
+
+    end
+
+  end
+
+end

data/lib/puree/xml_extractor/publication.rb

@@ -0,0 +1,189 @@
+module Puree
+
+  module XMLExtractor
+
+    # Publication XMLextractor.
+    #
+    class Publication < Puree::XMLExtractor::Resource
+
+      def initialize(xml:)
+        @resource_type = :publication
+        super
+      end
+
+      # @return [String, nil]
+      def category
+        xpath_query_for_single_value '/publicationCategory/publicationCategory/term/localizedString'
+      end
+
+      # @return [String, nil]
+      def description
+        xpath_query_for_single_value '/abstract/localizedString'
+      end
+
+      # @return [String, nil]
+      def doi
+        xpath_query_for_single_value '//doi'
+      end
+
+      # @return [Puree::Model::EventHeader, nil]
+      def event
+        xpath_result = xpath_query '/event'
+        if !xpath_result.empty?
+          header = Puree::Model::EventHeader.new
+          header.uuid = xpath_result.xpath('@uuid').text.strip
+          header.title = xpath_result.xpath('title/localizedString').text.strip
+          return header if header.data?
+        end
+        nil
+      end
+
+      # @return [Array<Puree::Model::File>]
+      def files
+        xpath_result = xpath_query '/electronicVersionAssociations/electronicVersionFileAssociations/electronicVersionFileAssociation/file'
+        docs = []
+        xpath_result.each do |d|
+          model = Puree::Model::File.new
+          model.name = d.xpath('fileName').text.strip
+          model.mime = d.xpath('mimeType').text.strip
+          model.size = d.xpath('size').text.strip.to_i
+          model.url = d.xpath('url').text.strip
+          docs << model
+        end
+        docs.uniq { |d| d.url }
+      end
+
+      # @return [Array<Puree::Model::OrganisationHeader>]
+      def organisations
+        xpath_result = xpath_query '/organisations/association/organisation'
+        Puree::XMLExtractor::Shared.organisation_multi_header xpath_result
+      end
+
+      # @return [Fixnum, nil]
+      def pages
+        xpath_query_for_single_value('/numberOfPages').to_i
+      end
+
+      # @return [Array<Puree::Model::EndeavourPerson>]
+      def persons_internal
+        persons 'internal'
+      end
+
+      # @return [Array<Puree::Model::EndeavourPerson>]
+      def persons_external
+        persons 'external'
+      end
+
+      # @return [Array<Puree::Model::EndeavourPerson>]
+      def persons_other
+        persons 'other'
+      end
+
+      # @return [Array<Puree::Model::PublicationStatus>]
+      def statuses
+        xpath_result = xpath_query '/publicationStatuses/publicationStatus'
+        data = []
+        xpath_result.each do |i|
+          s = Puree::Model::PublicationStatus.new
+          s.stage = i.xpath('publicationStatus/term/localizedString').text.strip
+
+          ymd = {}
+          ymd['year'] = i.xpath('publicationDate/year').text.strip
+          ymd['month'] = i.xpath('publicationDate/month').text.strip
+          ymd['day'] = i.xpath('publicationDate/day').text.strip
+
+          s.date = Puree::Util::Date.hash_to_time ymd
+
+          data << s
+        end
+        data.uniq { |d| d.stage }
+      end
+
+      # @return [String, nil]
+      def subtitle
+        xpath_query_for_single_value '/subtitle'
+      end
+
+      # @return [String, nil]
+      def title
+        xpath_query_for_single_value '/title'
+      end
+
+      # @return [String, nil]
+      def type
+        xpath_query_for_single_value '/typeClassification/term/localizedString'
+      end
+
+      private
+
+      # @return [Array<Endeavour::Person>]
+      def persons(type)
+        xpath_result = xpath_query '/persons/personAssociation'
+        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
+
+            role_uri = i.xpath('personRole/uri').text.strip
+            person.role = roles[role_uri]
+
+            arr << person if person.data?
+          end
+        end
+        arr.uniq { |d| d.uuid }
+      end
+
+      def roles
+        {
+            # Should build using '/dk/atira/pure/researchoutput/roles/' as prefix, with parameter
+
+            ## Article
+            '/dk/atira/pure/researchoutput/roles/contributiontojournal/author'          => 'Author',
+            '/dk/atira/pure/researchoutput/roles/contributiontojournal/illustrator'     => 'Illustrator',
+            '/dk/atira/pure/researchoutput/roles/contributiontojournal/editor'          => 'Editor',
+            '/dk/atira/pure/researchoutput/roles/contributiontojournal/translator'      => 'Translator',
+            '/dk/atira/pure/researchoutput/roles/contributiontojournal/publisher'       => 'Publisher',
+            '/dk/atira/pure/researchoutput/roles/contributiontojournal/guesteditor'     => 'Guest Editor',
+
+            ## Chapter
+            # Author
+            # Illustrator
+            # Editor
+            # Translator
+            # Publisher
+
+            # ... many, many more research output types ...
+
+            # Examples of others
+            '/dk/atira/pure/researchoutput/roles/bookanthology/author'                   => 'Author',
+            '/dk/atira/pure/researchoutput/roles/othercontribution/author'               => 'Author',
+            '/dk/atira/pure/researchoutput/roles/thesis/author'                          => 'Author',
+            '/dk/atira/pure/researchoutput/roles/workingpaper/author'                    => 'Author',
+            '/dk/atira/pure/researchoutput/roles/internalexternal/thesis/supervisor'     => 'Supervisor',
+            '/dk/atira/pure/researchoutput/roles/nontextual/artist'                      => 'Artist'
+        }
+      end
+
+    end
+
+  end
+
+end