puree 1.1.0 → 1.2.0

checksums.yaml

@@ -1,7 +1,7 @@
 ---
 SHA1:
-  metadata.gz: 51982259947535b8a9b88916eca1d629004b1d78
-  data.tar.gz: 0b4469e3b0f0881d45ebbdc7a90eb69f7b4e29cb
+  metadata.gz: 92900cf287eee283a00fc51d36aacdf62b6c0ef7
+  data.tar.gz: b3a7ea09d42501087a1bb86cd019d1ccbc6e0971
 SHA512:
-  metadata.gz: e1ab192064572d2935d3546577c4bdbe4dbbd624f9056f173da1d9d91f910beb493aa0dd17a13a1162003f618ea215f213643e6a3fc98509acdd89e8fa21b416
-  data.tar.gz: 3e94e0b3ae0f86fa0f91911e142d435d428c088735108d19aa5ca49c182d18f16602d83441d374158a9bbc8a4635d60c250bff43d4fc711634d4bd551e2ed0c7
+  metadata.gz: e0950129f4825a69f43286e5b6a933e0213e7ac48a99dfcd9cb8305d7c88f464b56472fad809d0b8eee68847d8b5b3decd16bdf4a911f3c23b037e51abff40f0
+  data.tar.gz: b61ec2735dc14cbdc196ad5049e4293f4f4972d5a5d0dac27d257af868ac20178be9e2e8682fd19bdcbbdafd7b60214bfecf9a0726cfe149454463f5d7fc9c4c

data/CHANGELOG.md

@@ -5,6 +5,19 @@ This project adheres to [Semantic Versioning](http://semver.org/).
 ## Unreleased
 - Metadata: activity?, clipping?, externalPerson?
 
+## 1.2.0 - 2017-03-31
+### Added
+- Publication - language, owner, translated_title, translated_subtitle.
+- Publication types - Doctoral Thesis and Master's Thesis.
+
+### Changed
+- Publication - common attribute set established. doi, event, pages become mixins
+for publication types.
+
+## 1.1.0 - 2017-03-24
+### Added
+- Publication - keywords, publisher.
+
 ## 1.0.0 - 2017-03-15
 ### Added
 - Dataset - legal conditions.

data/lib/puree.rb

@@ -4,6 +4,7 @@ require 'nokogiri'
 require 'puree/version'
 
 require 'puree/util/date'
+require 'puree/util/string'
 
 require 'puree/xml_extractor/shared'
 require 'puree/xml_extractor/base'
@@ -17,6 +18,12 @@ require 'puree/xml_extractor/organisation'
 require 'puree/xml_extractor/person'
 require 'puree/xml_extractor/project'
 require 'puree/xml_extractor/publication'
+require 'puree/xml_extractor/mixins/doi_mixin'
+require 'puree/xml_extractor/mixins/event_mixin'
+require 'puree/xml_extractor/mixins/pages_mixin'
+require 'puree/xml_extractor/thesis'
+require 'puree/xml_extractor/doctoral_thesis'
+require 'puree/xml_extractor/masters_thesis'
 require 'puree/xml_extractor/publisher'
 require 'puree/xml_extractor/server'
 
@@ -39,6 +46,9 @@ require 'puree/model/organisation'
 require 'puree/model/person'
 require 'puree/model/project'
 require 'puree/model/publication'
+require 'puree/model/thesis'
+require 'puree/model/doctoral_thesis'
+require 'puree/model/masters_thesis'
 require 'puree/model/publisher'
 require 'puree/model/related_content_header'
 require 'puree/model/spatial_point'
@@ -63,6 +73,9 @@ require 'puree/extractor/organisation'
 require 'puree/extractor/person'
 require 'puree/extractor/project'
 require 'puree/extractor/publication'
+require 'puree/extractor/thesis'
+require 'puree/extractor/doctoral_thesis'
+require 'puree/extractor/masters_thesis'
 require 'puree/extractor/publisher'
 require 'puree/extractor/collection'
 require 'puree/extractor/download'

data/lib/puree/extractor/collection.rb

@@ -19,10 +19,10 @@ module Puree
       #
       # @param limit [Fixnum]
       # @param offset [Fixnum]
-      # @param created_start [String]
-      # @param created_end [String]
-      # @param modified_start [String]
-      # @param modified_end [String]
+      # @param created_start [String] using format YYYY-MM-DD
+      # @param created_end [String] using format YYYY-MM-DD
+      # @param modified_start [String] using format YYYY-MM-DD
+      # @param modified_end [String] using format YYYY-MM-DD
       # @return [Array<Puree::Model::Resource subclass>] Resource metadata e.g. Puree::Model::Dataset.
       def get(
               limit:            0,

data/lib/puree/extractor/doctoral_thesis.rb

@@ -0,0 +1,24 @@
+module Puree
+  module Extractor
+
+    # Doctoral thesis extractor.
+    #
+    class DoctoralThesis < Puree::Extractor::Thesis
+
+      # @option (see Puree::Extractor::Resource#initialize)
+      def initialize(config)
+        set_model_type 'doctoral_thesis'
+        super
+      end
+
+      private
+
+      def combine_metadata
+        super
+        @model.type === 'Doctoral Thesis' ? @model : nil
+      end
+
+    end
+
+  end
+end

data/lib/puree/extractor/masters_thesis.rb

@@ -0,0 +1,24 @@
+module Puree
+  module Extractor
+
+    # Master's thesis extractor.
+    #
+    class MastersThesis < Puree::Extractor::Thesis
+
+      # @option (see Puree::Extractor::Resource#initialize)
+      def initialize(config)
+        set_model_type 'masters_thesis'
+        super
+      end
+
+      private
+
+      def combine_metadata
+        super
+        @model.type === "Master's Thesis" ? @model : nil
+      end
+
+    end
+
+  end
+end

data/lib/puree/extractor/publication.rb

@@ -16,14 +16,14 @@ module Puree
 
       def combine_metadata
         super
+
         @model.category = @extractor.category
         @model.description = @extractor.description
-        @model.doi = @extractor.doi
-        @model.event = @extractor.event
         @model.files = @extractor.files
         @model.keywords = @extractor.keywords
+        @model.language = @extractor.language
         @model.organisations = @extractor.organisations
-        @model.pages = @extractor.pages
+        @model.owner = @extractor.owner
         @model.persons_internal = @extractor.persons_internal
         @model.persons_external = @extractor.persons_external
         @model.persons_other = @extractor.persons_other
@@ -31,6 +31,8 @@ module Puree
         @model.statuses = @extractor.statuses
         @model.subtitle = @extractor.subtitle
         @model.title = @extractor.title
+        @model.translated_subtitle = @extractor.translated_subtitle
+        @model.translated_title = @extractor.translated_title
         @model.type = @extractor.type
         @model
       end

data/lib/puree/extractor/resource.rb

@@ -30,6 +30,11 @@ module Puree
 
       private
 
+      # For specialised extraction such as Publication subtypes e.g. doctoral_thesis
+      def set_model_type(type)
+        @model_type = type
+      end
+
       # Set content from XML.
       #
       # @param xml [String]
@@ -42,12 +47,20 @@ module Puree
 
       def setup(resource)
         @resource_type = resource
-        resource_class = "Puree::Model::#{resource.to_s.capitalize}"
+        if @model_type
+          resource_class = "Puree::Model::#{Puree::Util::String.titleize(@model_type.to_s)}"
+        else
+          resource_class = "Puree::Model::#{resource.to_s.capitalize}"
+        end
         @model = Object.const_get(resource_class).new
       end
 
       def make_xml_extractor xml
-        resource_class = "Puree::XMLExtractor::#{@resource_type.to_s.capitalize}"
+        if @model_type
+          resource_class = "Puree::XMLExtractor::#{Puree::Util::String.titleize(@model_type.to_s)}"
+        else
+          resource_class = "Puree::XMLExtractor::#{@resource_type.to_s.capitalize}"
+        end
         @extractor = Object.const_get(resource_class).new xml: xml
       end
 

data/lib/puree/extractor/thesis.rb

@@ -0,0 +1,34 @@
+module Puree
+  module Extractor
+
+    # Thesis extractor.
+    #
+    class Thesis < Puree::Extractor::Publication
+
+      # @option (see Puree::Extractor::Resource#initialize)
+      def initialize(config)
+        set_model_type 'thesis'
+        super
+      end
+
+      private
+
+      def combine_metadata
+        super
+
+        thesis_types = ["Master's Thesis", 'Doctoral Thesis']
+        return nil if !thesis_types.include? @model.type
+
+        @model.award_date = @extractor.award_date
+        @model.awarding_institution = @extractor.awarding_institution
+        @model.doi = @extractor.doi
+        @model.pages = @extractor.pages
+        @model.qualification = @extractor.qualification
+        @model.sponsors = @extractor.sponsors
+        @model
+      end
+
+    end
+
+  end
+end

data/lib/puree/model/doctoral_thesis.rb

@@ -0,0 +1,10 @@
+module Puree
+  module Model
+
+    # A doctoral thesis.
+    #
+    class DoctoralThesis < Puree::Model::Thesis
+
+    end
+  end
+end

data/lib/puree/model/masters_thesis.rb

@@ -0,0 +1,10 @@
+module Puree
+  module Model
+
+    # A Master's thesis.
+    #
+    class MastersThesis < Puree::Model::Thesis
+
+    end
+  end
+end

data/lib/puree/model/publication.rb

@@ -11,23 +11,20 @@ module Puree
       # @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<String>]
       attr_accessor :keywords
 
+      # @return [String, nil]
+      attr_accessor :language
+
       # @return [Array<Puree::Model::OrganisationHeader>]
       attr_accessor :organisations
 
-      # @return [Fixnum, nil]
-      attr_accessor :pages
+      # @return [Puree::Model::OrganisationHeader, nil]
+      attr_accessor :owner
 
       # @return [Array<Puree::Model::EndeavourPerson>]
       attr_accessor :persons_internal
@@ -50,6 +47,12 @@ module Puree
       # @return [String, nil]
       attr_accessor :title
 
+      # @return [String, nil]
+      attr_accessor :translated_subtitle
+
+      # @return [String, nil]
+      attr_accessor :translated_title
+
       # @return [String, nil]
       attr_accessor :type
 

data/lib/puree/model/thesis.rb

@@ -0,0 +1,28 @@
+module Puree
+  module Model
+
+    # A thesis.
+    #
+    class Thesis < Puree::Model::Publication
+
+      # @return [Time, nil]
+      attr_accessor :award_date
+
+      # @return [Puree::Model::OrganisationHeader, nil]
+      attr_accessor :awarding_institution
+
+      # @return [String, nil]
+      attr_accessor :doi
+
+      # # @return [Fixnum, nil]
+      attr_accessor :pages
+
+      # @return [String, nil]
+      attr_accessor :qualification
+
+      # @return [Array<String>, nil]
+      attr_accessor :sponsors
+
+    end
+  end
+end

data/lib/puree/util/date.rb

@@ -1,8 +1,7 @@
 module Puree
-
   module Util
 
-    # Date transformations
+    # Date transformations.
     #
     module Date
 
@@ -82,5 +81,4 @@ module Puree
     end
 
   end
-
 end

data/lib/puree/util/string.rb

@@ -0,0 +1,19 @@
+module Puree
+  module Util
+
+    # String utilities.
+    module String
+
+      # Titleize
+      # Assumes underscore as separator.
+      # e.g. foo_bar becomes FooBar
+      #
+      def self.titleize(x)
+        arr = "#{x}".split('_')
+        caps = arr.map { |i| i.capitalize }
+        caps.join
+      end
+
+    end
+  end
+end

data/lib/puree/version.rb

@@ -1,5 +1,5 @@
 module Puree
   # Semantic version number
   #
-  VERSION = "1.1.0"
+  VERSION = "1.2.0"
 end

data/lib/puree/xml_extractor/doctoral_thesis.rb

@@ -0,0 +1,14 @@
+module Puree
+  module XMLExtractor
+
+    # Doctoral thesis XML extractor.
+    #
+    class DoctoralThesis < Puree::XMLExtractor::Thesis
+
+      def initialize(xml:)
+        super
+      end
+
+    end
+  end
+end

data/lib/puree/xml_extractor/masters_thesis.rb

@@ -0,0 +1,14 @@
+module Puree
+  module XMLExtractor
+
+    # Master's thesis XML extractor.
+    #
+    class MastersThesis < Puree::XMLExtractor::Thesis
+
+      def initialize(xml:)
+        super
+      end
+
+    end
+  end
+end

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

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

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

@@ -0,0 +1,24 @@
+module Puree
+
+  module XMLExtractor
+
+    # Event extractor mixin.
+    #
+    module EventMixin
+
+      # @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
+
+    end
+
+  end
+end

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

@@ -0,0 +1,18 @@
+module Puree
+
+  module XMLExtractor
+
+    # Pages extractor mixin.
+    #
+    module PagesMixin
+
+      # @return [Fixnum, nil]
+      def pages
+        xpath_result = xpath_query_for_single_value('/numberOfPages')
+        xpath_result ? xpath_result.to_i : nil
+      end
+
+    end
+
+  end
+end

data/lib/puree/xml_extractor/publication.rb

@@ -2,7 +2,7 @@ module Puree
 
   module XMLExtractor
 
-    # Publication XMLextractor.
+    # Publication XML extractor.
     #
     class Publication < Puree::XMLExtractor::Resource
 
@@ -21,23 +21,6 @@ module Puree
         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'
@@ -67,16 +50,21 @@ module Puree
         data_arr.uniq
       end
 
+      # @return [String, nil]
+      def language
+        xpath_query_for_single_value '/language/term/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 [Fixnum, nil]
-      def pages
-        xpath_result = xpath_query_for_single_value('/numberOfPages')
-        xpath_result ? xpath_result.to_i : nil
+      # @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>]
@@ -129,6 +117,16 @@ module Puree
         xpath_query_for_single_value '/title'
       end
 
+      # @return [String, nil]
+      def translated_subtitle
+        xpath_query_for_single_value '/translatedSubtitle/localizedString'
+      end
+
+      # @return [String, nil]
+      def  translated_title
+        xpath_query_for_single_value '/translatedTitle/localizedString'
+      end
+
       # @return [String, nil]
       def type
         xpath_query_for_single_value '/typeClassification/term/localizedString'
@@ -187,7 +185,11 @@ module Puree
             # Author
             # Illustrator
             # Editor
-            # Translator
+            # Translator      # # @return [Fixnum, nil]
+      # def pages
+      #   xpath_result = xpath_query_for_single_value('/numberOfPages')
+      #   xpath_result ? xpath_result.to_i : nil
+      # end
             # Publisher
 
             # ... many, many more research output types ...

data/lib/puree/xml_extractor/thesis.rb

@@ -0,0 +1,44 @@
+module Puree
+  module XMLExtractor
+
+    # Thesis XML extractor.
+    #
+    class Thesis < Puree::XMLExtractor::Publication
+      include Puree::XMLExtractor::DoiMixin
+      include Puree::XMLExtractor::PagesMixin
+
+      def initialize(xml:)
+        super
+      end
+
+      # @return [Time, nil]
+      def award_date
+        xpath_result = xpath_query_for_single_value('/awardDate')
+        Time.parse xpath_result if xpath_result
+      end
+
+      # @return [Puree::Model::OrganisationHeader, nil]
+      def awarding_institution
+        xpath_result = xpath_query '/awardingInstitution/internalExternalOrganisationAssociation/organisation'
+        Puree::XMLExtractor::Shared.organisation_header xpath_result
+      end
+
+      # @return [String, nil]
+      def qualification
+        types = {
+            '/dk/atira/pure/thesis/qualification/mphil' => 'MPhil',
+            '/dk/atira/pure/thesis/qualification/phd' => 'PhD',
+            '/dk/atira/pure/thesis/qualification/masters_by_research' => 'Masters by Research'
+        }
+        xpath_result = xpath_query_for_single_value '/qualification/uri'
+        types[xpath_result]
+      end
+
+      # @return [Array<String>]
+      def sponsors
+        xpath_query_for_multi_value '/sponsors/externalOrganisation/name'
+      end
+
+    end
+  end
+end

data/spec/spec_helper.rb

@@ -4,6 +4,7 @@ require 'nokogiri'
 require 'puree/version'
 
 require 'puree/util/date'
+require 'puree/util/string'
 
 require 'puree/xml_extractor/shared'
 require 'puree/xml_extractor/base'

metadata

@@ -1,14 +1,14 @@
 --- !ruby/object:Gem::Specification
 name: puree
 version: !ruby/object:Gem::Version
-  version: 1.1.0
+  version: 1.2.0
 platform: ruby
 authors:
 - Adrian Albin-Clark
 autorequire: 
 bindir: bin
 cert_chain: []
-date: 2017-03-22 00:00:00.000000000 Z
+date: 2017-03-31 00:00:00.000000000 Z
 dependencies:
 - !ruby/object:Gem::Dependency
   name: http
@@ -73,10 +73,12 @@ files:
 - lib/puree/api/request.rb
 - lib/puree/extractor/collection.rb
 - lib/puree/extractor/dataset.rb
+- lib/puree/extractor/doctoral_thesis.rb
 - lib/puree/extractor/download.rb
 - lib/puree/extractor/event.rb
 - lib/puree/extractor/extractor.rb
 - lib/puree/extractor/journal.rb
+- lib/puree/extractor/masters_thesis.rb
 - lib/puree/extractor/organisation.rb
 - lib/puree/extractor/person.rb
 - lib/puree/extractor/project.rb
@@ -84,9 +86,11 @@ files:
 - lib/puree/extractor/publisher.rb
 - lib/puree/extractor/resource.rb
 - lib/puree/extractor/server.rb
+- lib/puree/extractor/thesis.rb
 - lib/puree/model/address.rb
 - lib/puree/model/copyright_license.rb
 - lib/puree/model/dataset.rb
+- lib/puree/model/doctoral_thesis.rb
 - lib/puree/model/download_header.rb
 - lib/puree/model/endeavour_person.rb
 - lib/puree/model/event.rb
@@ -96,6 +100,7 @@ files:
 - lib/puree/model/journal.rb
 - lib/puree/model/legal_condition.rb
 - lib/puree/model/link.rb
+- lib/puree/model/masters_thesis.rb
 - lib/puree/model/model.rb
 - lib/puree/model/organisation.rb
 - lib/puree/model/organisation_header.rb
@@ -111,15 +116,22 @@ files:
 - lib/puree/model/spatial_point.rb
 - lib/puree/model/structure.rb
 - lib/puree/model/temporal_range.rb
+- lib/puree/model/thesis.rb
 - lib/puree/util/date.rb
+- lib/puree/util/string.rb
 - lib/puree/util/util.rb
 - lib/puree/version.rb
 - lib/puree/xml_extractor/base.rb
 - lib/puree/xml_extractor/collection.rb
 - lib/puree/xml_extractor/dataset.rb
+- lib/puree/xml_extractor/doctoral_thesis.rb
 - lib/puree/xml_extractor/download.rb
 - lib/puree/xml_extractor/event.rb
 - lib/puree/xml_extractor/journal.rb
+- lib/puree/xml_extractor/masters_thesis.rb
+- lib/puree/xml_extractor/mixins/doi_mixin.rb
+- lib/puree/xml_extractor/mixins/event_mixin.rb
+- lib/puree/xml_extractor/mixins/pages_mixin.rb
 - lib/puree/xml_extractor/organisation.rb
 - lib/puree/xml_extractor/person.rb
 - lib/puree/xml_extractor/project.rb
@@ -128,6 +140,7 @@ files:
 - lib/puree/xml_extractor/resource.rb
 - lib/puree/xml_extractor/server.rb
 - lib/puree/xml_extractor/shared.rb
+- lib/puree/xml_extractor/thesis.rb
 - lib/puree/xml_extractor/xml_extractor.rb
 - puree.gemspec
 - spec/download_http_spec.rb