puree 2.7.0 → 2.8.0

checksums.yaml

@@ -1,7 +1,7 @@
 ---
 SHA256:
-  metadata.gz: d9e2d38e9e15a3b7da9abc9cc7350783a2bb4953b580ebbc4316532e10cff8a0
-  data.tar.gz: 68549513f64206b24c00fc0624584e27af33d98d95d33022c7af83c08cf7b5c8
+  metadata.gz: 0b91d7f85a6432c1233be1361c96669c94d88b17c32d5399bc8fcd0fc55c2c9b
+  data.tar.gz: 55eb39c43d50a7cf50e5297099ddcee1cd3471fa3679e3703063bfc6cb3215d1
 SHA512:
-  metadata.gz: 47858f41df3685614811f02d9336c2b0c19c8f88d3e14b797bec0a1154f3d1de73dd9ea7e42aeb31f2dfd20105b8bb9f9c120be5081b94ef61ff832449878b5f
-  data.tar.gz: 56dec4190827a75dc549af1cd7424146090efccbce1488a4996a585c0840e2d57c91196756f56227b598a1b4fa8b1cccee5fcf5c9388311f7643c4eb9d0ce685
+  metadata.gz: 91f7b09a0beaf0dc2badaea61ae558703ade237e906ec964a85ccec6bb8af45ed1ede6f8dd3fde74323d34749fbffa9dac7a02e8c1893c56eaacc970394fd377
+  data.tar.gz: 3102b9f34b06fa8ec08afd3dc47a1f0b914c0bbb6193fa7ce6f75c258e10f0454bb0cc6ada40fa60871da66f2adbf5f5ecc094b50a544bd9671415759f0a1250

data/CHANGELOG.md

@@ -2,6 +2,13 @@
 All notable changes to this project will be documented in this file.
 This project adheres to [Semantic Versioning](http://semver.org/).
 
+## 2.8.0 2019-07-04
+### Changed
+- For Pure API 514.
+
+### Added
+- Person - other_names.
+
 ## 2.7.0 2019-06-06
 ### Changed
 - For Pure API 513.

data/README.md

@@ -40,6 +40,7 @@ Purée version | Pure API version
 &gt;= 2.0, < 2.5 | 59, 510
 &gt;= 2.5, < 2.7 | 511, 512
 2.7 | 513
+2.8 | 514
 
 ## Extractor module
 ```ruby

data/lib/puree/model/person.rb

@@ -26,6 +26,8 @@ module Puree
       # @return [String, nil]
       attr_accessor :orcid
 
+      # @return [Array<Model::PersonName>]
+      attr_accessor :other_names
     end
   end
 end

data/lib/puree/version.rb

@@ -1,5 +1,5 @@
 module Puree
   # Semantic version number
   #
-  VERSION = "2.7.0"
+  VERSION = "2.8.0"
 end

data/lib/puree/xml_extractor/base.rb

@@ -14,8 +14,8 @@ module Puree
       #
       # @return [String, nil]
       def xpath_query_for_single_value(path)
-        xpath_result = xpath_query(path).text.strip
-        xpath_result.empty? ? nil : xpath_result
+        xpath_result = xpath_query(path)
+        xpath_result.empty? ? nil : xpath_result.first.text.strip
       end
 
       # XPath search for multiple values, at a given path.

data/lib/puree/xml_extractor/collection.rb

@@ -75,7 +75,7 @@ module Puree
       # @param xml [String]
       # @return [Hash{Symbol => Array<Puree::Model::ResearchOutput class/subclass>}]
       def self.research_outputs(xml)
-        path_from_root = File.join 'result', '/*'
+        path_from_root = File.join 'result/items', '/*'
         doc = Nokogiri::XML xml
         doc.remove_namespaces!
         xpath_result = doc.xpath path_from_root
@@ -86,8 +86,9 @@ module Puree
             other: []
         }
         xpath_result.each do |research_output|
-          type = research_output.xpath('type').text.strip
-          unless type.empty?
+          xpath_result_type = research_output.xpath('types/type')
+          type = xpath_result_type.first.text.strip unless xpath_result_type.empty?
+          if type
             case type
               when 'Journal article'
                 extractor = Puree::XMLExtractor::JournalArticle.new research_output.to_s
@@ -127,7 +128,7 @@ module Puree
       def self.models(resource_type, xml, xpath_root)
         doc = Nokogiri::XML xml
         doc.remove_namespaces!
-        path_from_root = File.join 'result', xpath_root
+        path_from_root = File.join 'result/items', xpath_root
         xpath_result = doc.xpath path_from_root
         data = []
         xpath_result.each do |i|

data/lib/puree/xml_extractor/dataset.rb

@@ -46,8 +46,8 @@ module Puree
           # 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?
+          document_license = d.xpath('documentLicenses/documentLicense').first
+          if document_license
             license = Puree::Model::CopyrightLicense.new
             license.name = document_license.text.strip
             # license.name = document_license.xpath('term/localizedString').text.strip
@@ -88,7 +88,7 @@ module Puree
       # @return [Array<String>]
       def spatial_places
         # Data from free-form text box
-        xpath_result = xpath_query '/geographicalCoverage'
+        xpath_result = xpath_query '/geographicalCoverages/geographicalCoverage'
         data = []
         xpath_result.each do |i|
           data << i.text.strip

data/lib/puree/xml_extractor/external_organisation.rb

@@ -14,7 +14,7 @@ module Puree
 
       # @return [String, nil]
       def name
-        xpath_query_for_single_value '/name'
+        xpath_query_for_single_value '/names/name'
       end
 
       private

data/lib/puree/xml_extractor/journal_article.rb

@@ -26,7 +26,8 @@ module Puree
           header = Puree::Model::JournalHeader.new
           header.title = xpath_result.xpath('title').text.strip
           journal = xpath_result.xpath('journal')
-          header.type = journal.xpath('type').text.strip
+          xpath_result_type = journal.xpath('types/type')
+          header.type = xpath_result_type.first.text.strip unless xpath_result_type.empty?
           header.uuid = journal.attr('uuid').text.strip
           header
         end

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

@@ -8,7 +8,7 @@ module Puree
 
       # @return [String, nil]
       def description
-        xpath_query_for_single_value '/abstract'
+        xpath_query_for_single_value '/abstracts/abstract'
       end
 
     end

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

@@ -8,7 +8,7 @@ module Puree
 
       # @return [String, nil]
       def description
-        xpath_query_for_single_value '/description'
+        xpath_query_for_single_value '/descriptions/description'
       end
 
     end

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

@@ -12,7 +12,8 @@ module Puree
         if !xpath_result.empty?
           header = Puree::Model::EventHeader.new
           header.uuid = xpath_result.xpath('@uuid').text.strip
-          header.title = xpath_result.xpath('name').text.strip
+          xpath_result_name = xpath_result.xpath('names/name')
+          header.title = xpath_result_name.first.text.strip unless xpath_result_name.empty?
           return header if header.data?
         end
         nil

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

@@ -32,7 +32,8 @@ module Puree
             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
+            xpath_result_role = i.xpath('personRoles/personRole')
+            person.role = xpath_result_role.first.text.strip unless xpath_result_role.empty?
             arr << person if person.data?
           end
         end

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

@@ -10,13 +10,14 @@ module Puree
       # @return [Array<Puree::Model::RelatedContentHeader>]
       def projects
         xpath_result = xpath_query '/relatedProjects/relatedProject'
-
         data_arr = []
         xpath_result.each { |i|
           related = Puree::Model::RelatedContentHeader.new
-          related.type = i.xpath('type').text.strip
-          related.title = i.xpath('name').text.strip
           related.uuid = i.attr('uuid').strip
+          xpath_result_name = i.xpath('names/name')
+          related.title = xpath_result_name.first.text.strip unless xpath_result_name.empty?
+          xpath_result_type = i.xpath('types/type')
+          related.type = xpath_result_type.first.text.strip unless xpath_result_type.empty?
           data_arr << related
         }
         data_arr.uniq { |d| d.uuid }

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

@@ -11,8 +11,10 @@ module Puree
         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
+        xpath_result_name = xpath_result.xpath('names/name')
+        h.name = xpath_result_name.first.text.strip unless xpath_result_name.empty?
+        xpath_result_type = xpath_result.xpath('types/type')
+        h.type = xpath_result_type.first.text.strip unless xpath_result_type.empty?
         h.data? ? h : nil
       end
 

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

@@ -13,9 +13,11 @@ module Puree
         data_arr = []
         xpath_result.each { |i|
           related = Puree::Model::RelatedContentHeader.new
-          related.type = i.xpath('type').text.strip
-          related.title = i.xpath('name').text.strip
           related.uuid = i.attr('uuid').strip
+          xpath_result_name = i.xpath('names/name')
+          related.title = xpath_result_name.first.text.strip unless xpath_result_name.empty?
+          xpath_result_type = i.xpath('types/type')
+          related.type = xpath_result_type.first.text.strip unless xpath_result_type.empty?
           data_arr << related
         }
         data_arr.uniq { |d| d.uuid }

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

@@ -8,7 +8,7 @@ module Puree
 
       # @return [String, nil]
       def title
-        xpath_query_for_single_value '/title'
+        xpath_query_for_single_value '/titles/title'
       end
 
     end

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

@@ -8,7 +8,7 @@ module Puree
 
       # @return [String, nil]
       def type
-        xpath_query_for_single_value '/type'
+        xpath_query_for_single_value '/types/type'
       end
 
     end

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

@@ -8,7 +8,7 @@ module Puree
 
       # @return [String, nil]
       def workflow
-        xpath_query_for_single_value '/workflow'
+        xpath_query_for_single_value '/workflows/workflow'
       end
 
     end

data/lib/puree/xml_extractor/organisational_unit.rb

@@ -25,8 +25,8 @@ module Puree
           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?
+          country = xpath_result.xpath('countries/country')
+          a.country = country.first.text.strip unless country.empty?
           a
         end
       end
@@ -38,7 +38,7 @@ module Puree
 
       # @return [String, nil]
       def name
-        xpath_query_for_single_value '/name'
+        xpath_query_for_single_value '/names/name'
       end
 
       # First parent

data/lib/puree/xml_extractor/person.rb

@@ -57,6 +57,21 @@ module Puree
         xpath_query_for_single_value '/orcid'
       end
 
+      # @return [Array<Model::PersonName>]
+      def other_names
+        xpath_result = xpath_query '/nameVariants/nameVariant/name'
+        data = []
+        xpath_result.each do |d|
+          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?
+          data << model
+        end
+        data.uniq
+      end
+
       private
 
       def xpath_root
@@ -72,6 +87,7 @@ module Puree
         @model.keywords = keywords
         @model.name = name
         @model.orcid = orcid
+        @model.other_names = other_names
         @model
       end      
 

data/lib/puree/xml_extractor/project.rb

@@ -50,7 +50,7 @@ module Puree
 
       # @return [String, nil]
       def status
-        xpath_query_for_single_value '/status'
+        xpath_query_for_single_value '/statuses/status'
       end
 
       # @return [Puree::Model::TemporalRange, nil]

data/lib/puree/xml_extractor/research_output.rb

@@ -23,12 +23,12 @@ module Puree
 
       # @return [String, nil]
       def bibliographical_note
-        xpath_query_for_single_value('/bibliographicalNote')
+        xpath_query_for_single_value '/bibliographicalNotes/bibliographicalNote'
       end
 
       # @return [String, nil]
       def category
-        xpath_query_for_single_value '/category'
+        xpath_query_for_single_value '/categories/category'
       end
 
       # Digital Object Identifier (first one, if many)
@@ -73,7 +73,7 @@ module Puree
 
       # @return [String, nil]
       def language
-        xpath_query_for_single_value '/language'
+        xpath_query_for_single_value '/languages/language'
       end
 
       # @return [Array<String>, nil]
@@ -83,7 +83,7 @@ module Puree
 
       # @return [String, nil]
       def open_access_permission
-        xpath_query_for_single_value '/openAccessPermission'
+        xpath_query_for_single_value '/openAccessPermissions/openAccessPermission'
       end
 
       # @return [Array<Puree::Model::EndeavourPerson>]
@@ -107,7 +107,7 @@ module Puree
         data = []
         xpath_result.each do |i|
           s = Puree::Model::PublicationStatus.new
-          s.stage = i.xpath('publicationStatus').text.strip
+          s.stage = i.xpath('publicationStatuses/publicationStatus').text.strip
 
           ymd = {}
           ymd['year'] = i.xpath('publicationDate/year').text.strip
@@ -156,12 +156,12 @@ module Puree
 
       # @return [String, nil]
       def translated_subtitle
-        xpath_query_for_single_value '/translatedSubTitle'
+        xpath_query_for_single_value '/translatedSubTitles/translatedSubTitle'
       end
 
       # @return [String, nil]
       def translated_title
-        xpath_query_for_single_value '/translatedTitle'
+        xpath_query_for_single_value '/translatedTitles/translatedTitle'
       end
 
       private

data/lib/puree/xml_extractor/shared.rb

@@ -10,8 +10,10 @@ module Puree
       def self.external_organisation_header(nokogiri_xml_element)
         h = Puree::Model::ExternalOrganisationHeader.new
         h.uuid = nokogiri_xml_element.xpath('@uuid').text.strip
-        h.name = nokogiri_xml_element.xpath('name').text.strip
-        h.type = nokogiri_xml_element.xpath('type').text.strip
+        xpath_result_name = nokogiri_xml_element.xpath('names/name')
+        h.name = xpath_result_name.first.text.strip unless xpath_result_name.empty?
+        xpath_result_type = nokogiri_xml_element.xpath('types/type')
+        h.type = xpath_result_type.first.text.strip unless xpath_result_type.empty?
         h.data? ? h : nil
       end
 
@@ -29,8 +31,10 @@ module Puree
       def self.organisation_header(nokogiri_xml_element)
         h = Puree::Model::OrganisationalUnitHeader.new
         h.uuid = nokogiri_xml_element.xpath('@uuid').text.strip
-        h.name = nokogiri_xml_element.xpath('name').text.strip
-        h.type = nokogiri_xml_element.xpath('type').text.strip
+        xpath_result_name = nokogiri_xml_element.xpath('names/name')
+        h.name = xpath_result_name.first.text.strip unless xpath_result_name.empty?
+        xpath_result_type = nokogiri_xml_element.xpath('types/type')
+        h.type = xpath_result_type.first.text.strip unless xpath_result_type.empty?
         h.data? ? h : nil
       end
 

data/lib/puree/xml_extractor/thesis.rb

@@ -27,7 +27,7 @@ module Puree
 
       # @return [String, nil]
       def qualification
-        xpath_query_for_single_value '/qualification'
+        xpath_query_for_single_value('/qualifications/qualification')
       end
 
       # @return [Array<Puree::Model::ExternalOrganisationHeader>]

data/puree.gemspec

@@ -10,8 +10,8 @@ Gem::Specification.new do |spec|
   spec.email         = 'a.albin-clark@lancaster.ac.uk'
   spec.summary       = %q{Metadata extraction from the Pure Research Information System.}
   spec.metadata = {
-    'source_code_uri' => 'https://github.com/lulibrary/puree',
-    "documentation_uri" => "https://www.rubydoc.info/gems/puree/2.7.0",
+    'source_code_uri' => "https://github.com/lulibrary/#{spec.name}",
+    "documentation_uri" => "https://www.rubydoc.info/gems/#{spec.name}/#{spec.version}",
   }
   spec.license       = 'MIT'
   spec.files         = `git ls-files -z`.split("\x0")

data/test/xml_extractor/xml_extractor_person_test.rb

@@ -51,6 +51,10 @@ class TestXMLExtractorPerson < Minitest::Test
 
     assert_instance_of String, x.orcid
     refute_empty x.orcid
+
+    assert_instance_of Array, x.other_names
+    assert_instance_of Puree::Model::PersonName, x.other_names.first
+    assert x.other_names.first.data?
   end
 
   # def test_scopus_id
@@ -80,6 +84,9 @@ class TestXMLExtractorPerson < Minitest::Test
     assert_nil x.name
 
     assert_nil x.orcid
+
+    assert_instance_of Array, x.other_names
+    assert_empty x.other_names
   end
 
   def test_model

data/test/xml_extractor/xml_extractor_research_output_test.rb

@@ -289,8 +289,12 @@ class TestXMLExtractorResearchOutput < Minitest::Test
     '<contributionToJournal>
       <relatedProjects>
         <relatedProject uuid="fe8aebdf-a926-4e7b-adf1-082425e50330">
-          <name>The Language Bases of Reading Comprehension</name>
-	   	    <type uri="/dk/atira/pure/upmproject/upmprojecttypes/upmproject/research">Research</type>
+          <names>
+            <name>The Language Bases of Reading Comprehension</name>
+          </names>
+          <types>
+            <type uri="/dk/atira/pure/upmproject/upmprojecttypes/upmproject/research">Research</type>
+          </types>
 	 	    </relatedProject>
 	    </relatedProjects>
     </contributionToJournal>'

metadata

@@ -1,14 +1,14 @@
 --- !ruby/object:Gem::Specification
 name: puree
 version: !ruby/object:Gem::Version
-  version: 2.7.0
+  version: 2.8.0
 platform: ruby
 authors:
 - Adrian Albin-Clark
 autorequire: 
 bindir: bin
 cert_chain: []
-date: 2019-06-06 00:00:00.000000000 Z
+date: 2019-07-31 00:00:00.000000000 Z
 dependencies:
 - !ruby/object:Gem::Dependency
   name: http
@@ -236,7 +236,7 @@ licenses:
 - MIT
 metadata:
   source_code_uri: https://github.com/lulibrary/puree
-  documentation_uri: https://www.rubydoc.info/gems/puree/2.7.0
+  documentation_uri: https://www.rubydoc.info/gems/puree/2.8.0
 post_install_message: 
 rdoc_options: []
 require_paths: