Merge pull request #585 from ndlib/DLTP-721-Support-Versions-for-OSF-…

…Archiv

Support Versions for OSF Archive

.started-issues

@@ -1,4 +1,3 @@
 478
 DLTP-638
 DLTP-717
-DLTP-764

app/assets/stylesheets/modules/attributes.css.scss

@@ -48,3 +48,13 @@
     }
   }
 }
+
+.table.related-to-works {
+  .selected-row {
+    border: 2px solid $blue-light;
+  }
+
+  .modifier.current {
+    width: 3em;
+  }
+}

app/mappers/as_jsonld_mapper.rb

@@ -31,6 +31,7 @@ def self.call(curation_concern, **keywords)
       'vracore'.to_sym => 'http://purl.org/vra/',
       'frels'.to_sym => 'info:fedora/fedora-system:def/relations-external#',
       'ms'.to_sym => 'http://www.ndltd.org/standards/metadata/etdms/1.1/',
+      'pav'.to_sym => 'http://purl.org/pav/',
       "fedora-model".to_sym => "info:fedora/fedora-system:def/model#",
       "hydra".to_sym => "http://projecthydra.org/ns/relations#",
       "hasModel".to_sym => {"@id" => "fedora-model:hasModel", "@type" => "@id" },
@@ -40,7 +41,7 @@ def self.call(curation_concern, **keywords)
       "isMemberOfCollection".to_sym => {"@id" => "frels:isMemberOfCollection", "@type" => "@id" },
       "isEditorOf".to_sym => {"@id" => "hydra:isEditorOf", "@type" => "@id" },
       "hasMember".to_sym => {"@id" => "frels:hasMember", "@type" => "@id" },
-      "previousVersion".to_sym => "http://purl.org/pav/previousVersion"
+      'previousVersion'.to_sym => 'http://purl.org/pav/previousVersion'
   }
 
   DEFAULT_ATTRIBUTES = {

app/repository_datastreams/osf_archive_datastream.rb

@@ -10,7 +10,13 @@ class OsfArchiveDatastream < ActiveFedora::NtriplesRDFDatastream
       index.as :stored_searchable
     end
 
-    map.source(to: 'source', in: RDF::DC)
+    map.source(to: 'source', in: RDF::DC) do |index|
+      index.as :stored_searchable
+    end
+
+    map.osf_project_identifier(to: 'osfProjectIdentifier', in: RDF::ND) do |index|
+      index.as :stored_searchable
+    end
 
     map.subject(to: 'subject', in: RDF::DC) do |index|
       index.as :stored_searchable

app/repository_models/osf_archive.rb

@@ -1,3 +1,16 @@
+# Models two "types" of OSF Archives:
+#
+# * OSF Project
+# * OSF Registration
+#
+# The OSF Registration is a "snapshot" of an OSF Project. The Registration has a URL
+# separate from the OSF Project; It points back to the OSF Project. In theory, if
+# we ingest the same OSF Registration we will have the same information; In practice,
+# some of the data for the registration may point to external sources that have changed
+# between ingests of that OSF Registration.
+#
+# The OSF Project is a living/mutable source. It represents the current state of the project.
+# When we ingest an OSF Project, that current state is captured.
 class OsfArchive < ActiveFedora::Base
   include ActiveModel::Validations
   include CurationConcern::Work
@@ -8,23 +21,64 @@ class OsfArchive < ActiveFedora::Base
   include ActiveFedora::RegisteredAttributes
   include CurationConcern::RemotelyIdentifiedByDoi::Attributes
   include CurationConcern::WithJsonMapper
-  
+
   before_validation :set_initial_values, on: :create
 
+  belongs_to :previousVersion, property: :previousVersion, class_name: "OsfArchive"
+
   class_attribute :human_readable_short_description
   self.human_readable_short_description = "Change me."
 
   def self.human_readable_type
     'OSF Archive'
   end
 
+  # These are included as convenience for developers.
+  SOLR_KEY_OSF_PROJECT_IDENTIFIER = 'desc_metadata__osf_project_identifier_tesim'.freeze
+  SOLR_KEY_ARCHIVED_DATE = 'desc_metadata__date_archived_dtsi'.freeze
+  SOLR_KEY_SOURCE = 'desc_metadata__source_tesim'.freeze
+
+  # Retrieve all archived versions of the source project (including registrations)
+  # in date_archived descending order.
+  #
+  # @return [Array<OsfArchive>]
+  # @see ./spec/repository_models/osf_archive_spec.rb
+  def archived_versions_of_source_project
+    @archived_versions_of_source_project ||= begin
+      conditions = { SOLR_KEY_OSF_PROJECT_IDENTIFIER => osf_project_identifier }
+      options = { sort: "#{SOLR_KEY_ARCHIVED_DATE} desc" }
+      solr_results = self.class.find_with_conditions(conditions, options)
+      ActiveFedora::SolrService.reify_solr_results(solr_results, load_from_solr: true)
+    end
+  end
+
   def set_initial_values
     self.date_modified = Date.today
-    self.date_archived = Date.today
-    self.type = human_readable_type
+    self.date_archived ||= Date.today
+    determine_type_and_osf_project_identifier
   end
   private :set_initial_values
 
+  def determine_type_and_osf_project_identifier
+    if osf_project_identifier.present?
+      if osf_project_identifier == osf_source_slug
+        self.type = "OSF Project"
+      else
+        self.type = "OSF Registration"
+      end
+    else
+      if osf_source_slug.present?
+        self.osf_project_identifier = osf_source_slug
+        self.type = "OSF Project"
+      end
+    end
+  end
+  private :determine_type_and_osf_project_identifier
+
+  def osf_source_slug
+    self.source.to_s.split('/').last
+  end
+
   has_metadata 'descMetadata', type: OsfArchiveDatastream
 
   attribute :affiliation,
@@ -41,14 +95,18 @@ def set_initial_values
 
   attribute :source,
     datastream: :descMetadata, multiple: false,
-    label: 'Original OSF Project',
+    label: 'Original OSF URL',
     validates: {
       format: {
       with: URI::regexp(%w(http https ftp)),
       message: 'must be a valid URL.'
       }
     }
 
+  attribute :osf_project_identifier,
+    datastream: :descMetadata, multiple: false,
+    label: 'OSF Project Identifier'
+
   attribute :subject,
     datastream: :descMetadata, multiple: true
 

app/views/curation_concern/osf_archives/_attributes.html.erb

@@ -9,7 +9,8 @@
     <%= curation_concern_attribute_to_html(curation_concern, :date_created, 'Date Created') %>
     <%= curation_concern_attribute_to_html(curation_concern, :subject, "Subject") %>
     <%= curation_concern_attribute_to_html(curation_concern, :date_archived, "Archive Date") %>
-    <%= curation_concern_attribute_to_html(curation_concern, :source, "Original OSF Project") %>
+    <%= curation_concern_attribute_to_html(curation_concern, :source, "Original OSF URL") %>
+    <%= curation_concern_attribute_to_html(curation_concern, :osf_project_identifier, "OSF Project Identifier") %>
     <%= curation_concern_attribute_to_html(curation_concern, :language, "Language") %>
     <%= decode_administrative_unit(curation_concern, :administrative_unit, "Departments and Units") %>
     <%= curation_concern_attribute_to_html(curation_concern, :library_collections, "Member of") %>
@@ -31,3 +32,29 @@
     <% end %>
   </tbody>
 </table>
+
+<h2>Version History</h2>
+<table class="table table-striped <%= dom_class(curation_concern) %> related-to-works with-headroom">
+  <caption class="table-heading">
+    <p>These are different snapshots of the same OSF Project.</p>
+  </caption>
+  <thead>
+  <tr>
+    <th>&nbsp;</th>
+    <th>CurateND Identifier</th>
+    <th>Date Archived</th>
+  </tr>
+  </thead>
+  <tbody>
+  <% curation_concern.archived_versions_of_source_project.each do |archived_version| %>
+    <% is_viewing_version = (archived_version.pid == curation_concern.pid) %>
+    <tr class="referenced_by_works attributes <%= 'selected-row' if is_viewing_version %>">
+      <td class="modifier current"><%= is_viewing_version ? 'Viewing' : '&nbsp;'.html_safe %></td>
+      <td class="attribute noid">
+        <%= link_to_unless(is_viewing_version, archived_version.noid, polymorphic_path([:curation_concern, archived_version])) %>
+      </td>
+      <td class="attribute date_archived"><%= content_tag :time, archived_version.date_archived.strftime('%Y-%m-%d'), datetime: archived_version.date_archived.iso8601 %></td>
+    </tr>
+  <% end %>
+  </tbody>
+</table>

app/views/curation_concern/osf_archives/_form_additional_information.html.erb

@@ -1,5 +1,6 @@
 <fieldset class="optional prompt">
   <legend>Additional Information</legend>
+  <%= f.input :osf_project_identifier, label: "OSF Project Identifier", input_html: { class: 'input-xlarge' } %>
   <%= f.input :affiliation,
               label: 'Academic Status',
               as: :select,

app/views/curation_concern/osf_archives/_form_required_information.html.erb

@@ -11,5 +11,5 @@
               },
               label: 'Description'
   %>
-  <%= f.input :source,               label: "Original OSF Project", input_html: {class: 'input-xlarge', type:'url'} %>
+  <%= f.input :source,               label: "Original OSF URL", input_html: {class: 'input-xlarge', type:'url'} %>
 </fieldset>

config/predicate_mappings.yml

@@ -5,6 +5,7 @@
   fedora-model: info:fedora/fedora-system:def/model#
   fedora-relations-model: info:fedora/fedora-system:def/relations-external#
   hydramata-rel: http://projecthydra.org/ns/relations#
+  pav: 'http://purl.org/pav/'
 
 # namespace mappings---
 # you can add specific mappings for your institution by providing the following:
@@ -49,6 +50,8 @@
     :is_deployment_of: isDeploymentOf
     :has_service: hasService
     :has_model: hasModel
+  http://purl.org/pav/:
+    :previousVersion: previousVersion
   http://www.openarchives.org/OAI/2.0/:
     :oai_item_id: itemID
   http://projecthydra.org/ns/relations#:

lib/rdf/nd.rb

@@ -2,5 +2,6 @@
 module RDF
   class ND < Vocabulary("https://library.nd.edu/ns/terms/")
     property :alephIdentifier
+    property :osfProjectIdentifier, comment: "The identifier of the source OSF Project. Extends dc:identifier."
   end
 end

lib/rdf/pav.rb

@@ -0,0 +1,5 @@
+module RDF
+  class PAV < Vocabulary("http://purl.org/pav/")
+    property :previousVersion
+  end
+end

script/one-off/DLTP-721-seed-data.rb

@@ -0,0 +1,16 @@
+attributes = {
+  osf_project_identifier: 'abcde',
+  source: 'https://osf.io/abcde',
+  title: 'Example OSF Project',
+  description: "Many Bothan's died to bring you this project",
+  visibility: Hydra::AccessControls::AccessRight::VISIBILITY_TEXT_VALUE_PUBLIC
+}
+
+previous_project = OsfArchive.create!(attributes.merge(date_archived: '2016-01-02'))
+registration = OsfArchive.create!(attributes.merge(date_archived: '2016-02-02', source: 'https://osf.io/12345'))
+project = OsfArchive.create!(attributes.merge(date_archived: '2016-03-02', previousVersion: previous_project))
+
+puts "Previous Project URL: http://localhost:3000/show/#{previous_project.noid}"
+puts "Registration URL: http://localhost:3000/show/#{registration.noid}"
+puts "Project URL: http://localhost:3000/show/#{project.noid}"
+`open http://localhost:3000/show/#{project.noid}`

spec/factories/osf_archives_factory.rb

@@ -12,7 +12,10 @@
     date_created { Date.today }
     date_modified { Date.today }
     date_archived { Date.today }
+    # the source and osf_project_identifier are related
     source { 'https://osf.io/xxxxx/' }
+    osf_project_identifier { 'xxxxx' }
+
     visibility Hydra::AccessControls::AccessRight::VISIBILITY_TEXT_VALUE_AUTHENTICATED
 
     before(:create) { |work, evaluator|

spec/repository_models/osf_archive_spec.rb

@@ -6,21 +6,95 @@
   it { should have_unique_field(:title) }
   it { should have_unique_field(:source) }
   it { should have_unique_field(:type) }
+  it { should have_unique_field(:osf_project_identifier) }
 
   it_behaves_like 'is_a_curation_concern_model'
   it_behaves_like 'with_access_rights'
   it_behaves_like 'can_be_a_member_of_library_collections'
   it_behaves_like 'is_embargoable'
   it_behaves_like 'with_json_mapper'
 
-  describe 'new archive' do
+  describe 'new archive (after validation)' do
     let(:archive) {OsfArchive.new}
 
-    it "should set initialize dates and stamp type on create" do
+    it "should set initialize dates" do
       archive.valid?
       expect(archive.date_modified).to eq(Date.today)
       expect(archive.date_archived).to eq(Date.today)
-      expect(archive.type).to eq('OSF Archive')
+    end
+
+    [
+      { attributes: { osf_project_identifier: "abcde", source: 'https://osf.io/abcde' }, type: 'OSF Project' },
+      { attributes: { osf_project_identifier: "12345", source: 'https://osf.io/abcde' }, type: 'OSF Registration' },
+      { attributes: { osf_project_identifier: "", source: 'https://osf.io/abcde' }, type: 'OSF Project' },
+      { attributes: { osf_project_identifier: "", source: "" }, type: nil },
+      { attributes: { osf_project_identifier: "12345", source: "" }, type: 'OSF Registration' }
+    ].each_with_index do |data, index|
+      it "will assign dc:type of #{data.fetch(:type).inspect} for attributes #{data.fetch(:attributes).inspect} (Scenario ##{index})" do
+        osf_archive = OsfArchive.new(data.fetch(:attributes))
+        osf_archive.valid? # required because we are setting via before validation
+        expect(osf_archive.type).to eq(data.fetch(:type))
+      end
+    end
+  end
+
+  describe 'related versions of OSF Archive objects' do
+    let(:osf_project_identifier) { 'abcde' }
+    let(:osf_registration_identifier) { '12345' }
+    let!(:previous_version) do
+      FactoryGirl.create(
+        :osf_archive,
+        source: "https://osf.io/#{osf_project_identifier}",
+        osf_project_identifier: osf_project_identifier,
+        date_archived: 2.days.ago
+      )
+    end
+    let!(:current_version) do
+      FactoryGirl.create(
+        :osf_archive,
+        previousVersion: previous_version,
+        source: "https://osf.io/#{osf_project_identifier}",
+        osf_project_identifier: osf_project_identifier,
+        date_archived: 1.days.ago
+      )
+    end
+    let!(:project_registration) do
+      FactoryGirl.create(
+        :osf_archive,
+        source: "https://osf.io/#{osf_registration_identifier}",
+        osf_project_identifier: osf_project_identifier,
+        date_archived: 0.days.ago
+      )
+    end
+
+    it 'has expected JSON-LD, RELS-EXT, setter/getter behavior, and #archived_versions_of_source_project' do
+      osf_archive = OsfArchive.find(current_version.id)
+
+      # Ensuring that the previousVersion works
+      expect(osf_archive.previousVersion).to eq(previous_version)
+
+      # Ensuring a well configured SOLR
+      to_solr = osf_archive.to_solr
+      expect(to_solr.fetch(OsfArchive::SOLR_KEY_SOURCE)).to eq([osf_archive.source])
+      expect(to_solr.fetch(OsfArchive::SOLR_KEY_OSF_PROJECT_IDENTIFIER)).to eq([osf_project_identifier])
+      expect(to_solr.key?(OsfArchive::SOLR_KEY_ARCHIVED_DATE)).to eq(true)
+
+      # Ensuring #archived_versions_of_source_project
+      archived_versions_of_source_project = osf_archive.archived_versions_of_source_project
+      expect(archived_versions_of_source_project).to eq([
+        previous_version, current_version, project_registration
+      ])
+
+      jsonld = osf_archive.as_jsonld
+      # Ensuring that the as_jsonld contains the correct relationship
+      expect(jsonld.fetch('@context').fetch('pav')).to eq('http://purl.org/pav/')
+      expect(jsonld.fetch('@context').fetch('previousVersion')).to eq('pav:previousVersion')
+      expect(jsonld.fetch("previousVersion")).to eq({"@id" => previous_version.pid})
+
+      # Ensuring that we have a meaningful RELS-EXT
+      rels_ext = osf_archive.datastreams.fetch('RELS-EXT').content
+      expect(rels_ext).to include(%(xmlns:pav="http://purl.org/pav/"))
+      expect(rels_ext).to include(%(<pav:previousVersion rdf:resource="info:fedora/#{previous_version.pid}"></pav:previousVersion>))
     end
   end
 end