puree 0.20.0 → 1.0.0

data/lib/puree/model/publication.rb

@@ -0,0 +1,53 @@
+module Puree
+  module Model
+
+    # A publication.
+    #
+    class Publication < Resource
+
+      # @return [String, nil]
+      attr_accessor :category
+
+      # @return [String, nil]
+      attr_accessor :description
+
+      # @return [String, nil]
+      attr_accessor :doi
+
+      # @return [Puree::Model::EventHeader, nil]
+      attr_accessor :event
+
+      # @return [Array<Puree::Model::File>]
+      attr_accessor :files
+
+      # @return [Array<Puree::Model::OrganisationHeader>]
+      attr_accessor :organisations
+
+      # @return [Fixnum, nil]
+      attr_accessor :pages
+
+      # @return [Array<Puree::Model::EndeavourPerson>]
+      attr_accessor :persons_internal
+
+      # @return [Array<Puree::Model::EndeavourPerson>]
+      attr_accessor :persons_external
+
+      # @return [Array<Puree::Model::EndeavourPerson>]
+      attr_accessor :persons_other
+
+      # @return [Array<Puree::Model::PublicationStatus>]
+      attr_accessor :statuses
+
+      # @return [String, nil]
+      attr_accessor :subtitle
+
+      # @return [String, nil]
+      attr_accessor :title
+
+      # @return [String, nil]
+      attr_accessor :type
+
+
+    end
+  end
+end

data/lib/puree/model/publication_status.rb

@@ -0,0 +1,21 @@
+module Puree
+  module Model
+
+    # Status of a publication.
+    #
+    class PublicationStatus < Puree::Model::Structure
+
+      # @return [String, nil]
+      attr_reader :stage
+
+      # @return [Time, nil]
+      attr_accessor :date
+
+      # @param v [String]
+      def stage=(v)
+        @stage = v if v && !v.empty?
+      end
+
+    end
+  end
+end

data/lib/puree/model/publisher.rb

@@ -0,0 +1,13 @@
+module Puree
+  module Model
+
+    # A publisher.
+    #
+    class Publisher < Resource
+
+      # @return [String, nil]
+      attr_accessor :name
+
+    end
+  end
+end

data/lib/puree/model/related_content_header.rb

@@ -0,0 +1,34 @@
+module Puree
+  module Model
+
+    # A minimal representation of related content.
+    #
+    class RelatedContentHeader < Puree::Model::Structure
+
+      # @return [String, nil]
+      attr_reader :uuid
+
+      # @return [String, nil]
+      attr_reader :title
+
+      # @return [String, nil]
+      attr_reader :type
+
+      # @param v [String]
+      def uuid=(v)
+        @uuid = v if v && !v.empty?
+      end
+
+      # @param v [String]
+      def title=(v)
+        @title = v if v && !v.empty?
+      end
+
+      # @param v [String]
+      def type=(v)
+        @type = v if v && !v.empty?
+      end
+
+    end
+  end
+end

data/lib/puree/model/resource.rb

@@ -0,0 +1,42 @@
+module Puree
+  module Model
+
+    # A resource.
+    #
+    class Resource
+
+      # @return [String, nil]
+      attr_reader :uuid
+
+      # @return [Time, nil]
+      attr_reader :created
+
+      # @return [Time, nil]
+      attr_reader :modified
+
+      # @return [String, nil]
+      attr_reader :locale
+
+      # @param v [String]
+      def uuid=(v)
+        @uuid = v if v && !v.empty?
+      end
+
+      # @param v [Time]
+      def created=(v)
+        @created = v
+      end
+
+      # @param v [Time]
+      def modified=(v)
+        @modified = v
+      end
+
+      # @param v [String]
+      def locale=(v)
+        @locale = v if v && !v.empty?
+      end
+
+    end
+  end
+end

data/lib/puree/model/server.rb

@@ -0,0 +1,13 @@
+module Puree
+  module Model
+
+    # Pure host information.
+    #
+    class Server
+
+    # @return [String, nil]
+    attr_accessor :version
+
+    end
+  end
+end

data/lib/puree/model/spatial_point.rb

@@ -0,0 +1,16 @@
+module Puree
+  module Model
+
+    # A geographical point.
+    #
+    class SpatialPoint < Puree::Model::Structure
+
+      # @return [Float, nil]
+      attr_accessor :latitude
+
+      # @return [Float, nil]
+      attr_accessor :longitude
+
+    end
+  end
+end

data/lib/puree/model/structure.rb

@@ -0,0 +1,18 @@
+module Puree
+  module Model
+
+    # A structure.
+    #
+    class Structure
+
+      def data?
+        found = false
+        instance_variables.each do |i|
+          found = true if i
+        end
+        found
+      end
+
+    end
+  end
+end

data/lib/puree/model/temporal_range.rb

@@ -0,0 +1,15 @@
+module Puree
+  module Model
+
+    # One or two dates which are part of a date range.
+    class TemporalRange < Puree::Model::Structure
+
+        # @return [Time, nil]
+        attr_accessor :start
+
+        # @return [Time, nil]
+        attr_accessor :end
+
+    end
+  end
+end

data/lib/puree/util/date.rb

@@ -0,0 +1,86 @@
+module Puree
+
+  module Util
+
+    # Date transformations
+    #
+    module Date
+
+      # Converts a date with variable components present (year, month, day) to Time format.
+      #
+      # @param data [Hash]
+      # @return [Time]
+      def self.hash_to_time(data)
+        h = normalise data
+        if !h['year'].empty? && !h['month'].empty? && !h['day'].empty?
+          return Time.new h['year'].to_i,
+                          h['month'].to_i,
+                          h['day'].to_i
+        end
+        if !h['year'].empty? && !h['month'].empty?
+          return Time.new h['year'].to_i,
+                          h['month'].to_i
+        end
+        if !h['year'].empty?
+          return Time.new h['year'].to_i
+        end
+        nil
+      end
+
+      # Converts a date with three components (year, month, day) to ISO 8601 date format.
+      #
+      # @param data [Hash]
+      # @return [String]
+      def self.iso(data)
+        iso_date = ''
+        year =  data['year']
+        month = data['month']
+        day = data['day']
+        if !year.empty?
+          iso_date << year
+        else
+          iso_date
+        end
+        if !month.empty?
+          # Add leading zero to convert to ISO 8601 date component
+          if month.length < 2
+            month.insert(0, '0')
+          end
+          iso_date << '-' + month
+        else
+          iso_date
+        end
+        if !day.empty?
+          # Add leading zero to convert to ISO 8601 date component
+          if day.length < 2
+            day.insert(0, '0')
+          end
+          iso_date << '-' + day
+        end
+        iso_date
+      end
+
+      # Forces a date to have three components (year, month, day).
+      #
+      # @param data [Hash]
+      # @return [Hash]
+      def self.normalise(data)
+        if !data.nil? && !data.empty?
+          date = {}
+          year =  data['year']
+          month = data['month']
+          day = data['day']
+          date['year'] = year ? year : ''
+          date['month'] = month ? month : ''
+          date['day'] = day ? day : ''
+          date
+        else
+          {}
+        end
+      end
+
+    end
+
+  end
+
+end

data/lib/puree/util/util.rb

@@ -0,0 +1,8 @@
+module Puree
+
+  # Utilities.
+  module Util
+
+  end
+
+end

data/lib/puree/version.rb

@@ -1,5 +1,5 @@
 module Puree
   # Semantic version number
   #
-  VERSION = "0.20.0"
+  VERSION = "1.0.0"
 end

data/lib/puree/xml_extractor/base.rb

@@ -0,0 +1,47 @@
+module Puree
+
+  module XMLExtractor
+
+    # Base XML extractor.
+    #
+    class Base
+
+      def initialize(xml:)
+        @api_map = Puree::API::Map.new.get
+        make_doc xml
+      end
+
+      # XPath search for a single value, at a given path.
+      #
+      # @return [String, nil]
+      def xpath_query_for_single_value(path)
+        xpath_result = xpath_query(path).text.strip
+        xpath_result.empty? ? nil : xpath_result
+      end
+
+      # XPath search for multiple values, at a given path.
+      #
+      # @return [Array<String>]
+      def xpath_query_for_multi_value(path)
+        xpath_result = xpath_query path
+        arr = []
+        xpath_result.each { |i| arr << i.text.strip }
+        arr.uniq
+      end
+
+      private
+
+      def make_doc(xml)
+        @doc = Nokogiri::XML xml
+        @doc.remove_namespaces!
+      end
+
+      def service_response_name
+        @api_map[:resource_type][@resource_type][:response]
+      end
+
+    end
+
+  end
+
+end

data/lib/puree/xml_extractor/collection.rb

@@ -0,0 +1,40 @@
+module Puree
+
+  module XMLExtractor
+
+    # Collection XML extractor.
+    #
+    class Collection < Puree::XMLExtractor::Base
+
+      def initialize(xml:)
+        super
+      end
+
+      # Records available in collection.
+      #
+      # @return [Fixnum]
+      def count
+        @doc.xpath('//count').text.strip.to_i
+      end
+
+      # Resource UUIDs in collection response.
+      #
+      # @return [Array<String>]
+      def uuids
+        arr = []
+        xpath_result = @doc.xpath '//renderedItem/@renderedContentUUID'
+        xpath_result.each { |i| arr << i.text.strip }
+        arr
+      end
+
+      # Is there any data after get?
+      # @return [Boolean]
+      def get_data?
+        count >= 1 ? true : false
+      end
+
+    end
+
+  end
+
+end

data/lib/puree/xml_extractor/dataset.rb

@@ -0,0 +1,305 @@
+module Puree
+
+  module XMLExtractor
+
+    # Dataset XML extractor.
+    #
+    class Dataset < Puree::XMLExtractor::Resource
+
+      def initialize(xml:)
+        super
+        @resource_type = :dataset
+      end
+
+      # Open access permission
+      # @return [String, nil]
+      def access
+        xpath_query_for_single_value '/openAccessPermission/term/localizedString'
+      end
+
+      # Combines projects and publications
+      # @return [Array<Puree::Model::RelatedContentHeader>]
+      def associated
+        xpath_result = xpath_query '/associatedContent/relatedContent'
+        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.uuid = i.attr('uuid').strip
+          data_arr << related
+        }
+        data_arr.uniq { |d| d.uuid }
+      end
+
+      # Date made available
+      # @return [Time, nil]
+      def available
+        Puree::Util::Date.hash_to_time temporal_date('dateMadeAvailable')
+      end
+
+      # @return [String, nil]
+      def description
+        xpath_query_for_single_value '/descriptions/classificationDefinedField/value/localizedString'
+      end
+
+      # Digital Object Identifier
+      # @return [String, nil]
+      def doi
+        xpath_query_for_single_value '/doi'
+      end
+
+      # Supporting files
+      # @return [Array<Puree::Model::File>]
+      def files
+        xpath_result = xpath_query '/documents/document'
+        docs = []
+        xpath_result.each do |d|
+          doc = Puree::Model::File.new
+          doc.name = d.xpath('fileName').text.strip
+          doc.mime = d.xpath('mimeType').text.strip
+          doc.size = d.xpath('size').text.strip.to_i
+          doc.url = d.xpath('url').text.strip
+          # doc['createdDate'] = d.xpath('createdDate').text.strip
+          # doc['visibleOnPortalDate'] = d.xpath('visibleOnPortalDate').text.strip
+          # doc['limitedVisibility'] = d.xpath('limitedVisibility').text.strip
+          document_license = d.xpath('documentLicense')
+          if !document_license.empty?
+            license = Puree::Model::CopyrightLicense.new
+            license.name = document_license.xpath('term/localizedString').text.strip
+            license.url = document_license.xpath('description/localizedString').text.strip
+            doc.license = license if license.data?
+          end
+          docs << doc
+        end
+        docs.uniq { |d| d.url }
+      end
+
+      # @return [Array<String>]
+      def keywords
+        xpath_result =  xpath_query '/keywordGroups/keywordGroup/keyword/userDefinedKeyword/freeKeyword'
+        data_arr = xpath_result.map { |i| i.text.strip }
+        data_arr.uniq
+      end
+
+      # @return [Array<Puree::Model::LegalCondition>]
+      def legal_conditions
+        xpath_result = xpath_query '/legalConditions/legalCondition'
+        data = []
+        xpath_result.each { |i|
+          model =  Puree::Model::LegalCondition.new
+          model.name = i.xpath('typeClassification/term/localizedString').text.strip
+          model.description = i.xpath('description').text.strip
+          data << model
+        }
+        data.uniq { |d| d.name }
+      end
+
+      # @return [Array<Puree::Model::Link>]
+      def links
+        xpath_result = xpath_query '/links/link'
+        data = []
+        xpath_result.each { |i|
+          model =  Puree::Model::Link.new
+          model.description = i.xpath('description').text.strip
+          model.url = i.xpath('url').text.strip
+          data << model
+        }
+        data.uniq { |d| d.url }
+      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 owner
+        xpath_result = xpath_query '/managedBy'
+        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
+
+      # Date of data production
+      # @return [Puree::Model::TemporalRange, nil]
+      def production
+        temporal_range 'dateOfDataProduction', 'endDateOfDataProduction'
+      end
+
+      # @return [Array<Puree::Model::RelatedContentHeader>]
+      def projects
+        associated_type('Research').uniq
+      end
+
+      # @return [Array<Puree::Model::RelatedContentHeader>]
+      def publications
+        data_arr = []
+        associated.each do |i|
+          if i.type != 'Research'
+            data_arr << i
+          end
+        end
+        data_arr
+      end
+
+      # @return [String, nil]
+      def publisher
+        xpath_query_for_single_value '/publisher/name'
+      end
+
+      # @return [Array<String>]
+      def spatial_places
+        # Data from free-form text box
+        xpath_result = xpath_query '/geographicalCoverage/localizedString'
+        data = []
+        xpath_result.each do |i|
+          data << i.text.strip
+        end
+        data.uniq
+      end
+
+      # Spatial coverage point
+      # @return [Puree::Model::SpatialPoint, nil]
+      def spatial_point
+        xpath_result = xpath_query '/geoLocation/point'
+        point = Puree::Model::SpatialPoint.new
+        if !xpath_result[0].nil?
+          arr = xpath_result.text.split(',')
+          point.latitude = arr[0].strip.to_f
+          point.longitude = arr[1].strip.to_f
+          point
+        end
+        nil
+      end
+
+      # def state
+      #   # useful?
+      #   /startedWorkflow/state
+      # end
+
+      # Temporal coverage
+      # @return [Puree::Model::TemporalRange, nil]
+      def temporal
+        temporal_range 'temporalCoverageStartDate', 'temporalCoverageEndDate'
+      end
+
+      # @return [String, nil]
+      def title
+        xpath_query_for_single_value '/title/localizedString'
+      end
+
+      private
+
+      def associated_type(type)
+        data_arr = []
+        associated.each do |i|
+          if i.type === type
+            related = Puree::Model::RelatedContentHeader.new
+            related.type = i.type
+            related.title = i.title
+            related.uuid = i.uuid
+            data_arr << related
+          end
+        end
+        data_arr.uniq
+      end
+
+      # @return [Array<Endeavour::Person>]
+      def persons(type)
+        xpath_result = xpath_query '/persons/dataSetPersonAssociation'
+        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
+
+      # Temporal range
+      # @return [Puree::Model::TemporalRange, nil]
+      def temporal_range(start_path, end_path)
+        range_start = Puree::Util::Date.hash_to_time temporal_date(start_path)
+        range_end = Puree::Util::Date.hash_to_time temporal_date(end_path)
+        if range_start || range_end
+          range = Puree::Model::TemporalRange.new
+          range.start = range_start if range_start
+          range.end = range_end if range_end
+          return range
+        end
+        nil
+      end
+
+      # Temporal coverage date
+      # @return [Hash]
+      def temporal_date(date_path)
+        path = "/#{date_path}"
+        xpath_result = xpath_query path
+        o = {}
+        o['year'] = xpath_result.xpath('year').text.strip
+        o['month'] = xpath_result.xpath('month').text.strip
+        o['day'] = xpath_result.xpath('day').text.strip
+        Puree::Util::Date.normalise o
+      end
+
+      def roles
+        {
+            '/dk/atira/pure/dataset/roles/dataset/contributor'    => 'Contributor',
+            '/dk/atira/pure/dataset/roles/dataset/creator'        => 'Creator',
+            '/dk/atira/pure/dataset/roles/dataset/datacollector'  => 'Data Collector',
+            '/dk/atira/pure/dataset/roles/dataset/datamanager'    => 'Data Manager',
+            '/dk/atira/pure/dataset/roles/dataset/distributor'    => 'Distributor',
+            '/dk/atira/pure/dataset/roles/dataset/editor'         => 'Editor',
+            '/dk/atira/pure/dataset/roles/dataset/funder'         => 'Funder',
+            '/dk/atira/pure/dataset/roles/dataset/owner'          => 'Owner',
+            '/dk/atira/pure/dataset/roles/dataset/other'          => 'Other',
+            '/dk/atira/pure/dataset/roles/dataset/producer'       => 'Producer',
+            '/dk/atira/pure/dataset/roles/dataset/rightsholder'   => 'Rights Holder',
+            '/dk/atira/pure/dataset/roles/dataset/sponsor'        => 'Sponsor',
+            '/dk/atira/pure/dataset/roles/dataset/supervisor'     => 'Supervisor'
+        }
+      end
+
+    end
+
+  end
+
+end