ddr-models 3.0.0.beta.17 → 3.0.0.beta.18

checksums.yaml

@@ -1,7 +1,7 @@
 ---
 SHA1:
-  metadata.gz: 5ada8e4f7256516690e6fc879f9c7a76ff85d254
-  data.tar.gz: f6de430ff6b09be6010d34456a1e5a4e5ee59f86
+  metadata.gz: 16f19bb0a9c0d7569797de024db2638a3e1b4e2a
+  data.tar.gz: 1e83fcc5c343aeed276d1a36de4d99f9b95dc125
 SHA512:
-  metadata.gz: 95c1a3beb7903628cfabe5968fa3ac91b1101fbdb3ab0a35ab6a7a67b79bd394e28d91a6758983b4d44fdba83bcf168f6a3a027f3b3a9facc5e299c29b4c99ae
-  data.tar.gz: cc7cb67e7e46437ad4d1c7f48dfcd673de6673b3d37bd1edbdb3e6459df0c5d14d29d65bdb481b6f4afb0c43445297ebe6298298327fda9f11f86e9c4acd32fd
+  metadata.gz: 4abaf0450a6f928fc6c3c52da7c4a570028c82dcd58fedeff07bf24262bdce23d6a4b6c7ee30f49894120c52381d6d93c174515f2ea49ec726ae8ac8de347c34
+  data.tar.gz: 6975b9bf6181df4407f480b44f21f4d697fef2c989fc2f878c63c1d524f247210239042e751572530028af6b2deab2bc9c4f8827cc5054d777ef1f38423b0511

data/app/models/component.rb

@@ -17,6 +17,8 @@ class Component < Ddr::Models::Base
              predicate: ::RDF::URI("http://www.loc.gov/mix/v20/externalTarget#hasExternalTarget"),
              class_name: "Target"
 
+  after_save :index_parent, if: :has_extracted_text?, unless: "parent.nil?"
+
   def collection
     self.parent.parent rescue nil
   end
@@ -29,4 +31,8 @@ class Component < Ddr::Models::Base
     parent.present? && parent.published?
   end
 
+  def index_parent
+    Resque.enqueue(Ddr::Jobs::UpdateIndex, parent_id)
+  end
+
 end

data/app/models/item.rb

@@ -21,4 +21,17 @@ class Item < Ddr::Models::Base
     parent.present? && parent.published?
   end
 
+  def children_having_extracted_text
+    item = self
+    Ddr::Index::Query.new do
+      is_part_of item
+      where attached_files_having_content: "extractedText"
+      fields :id, :extracted_text
+    end
+  end
+
+  def all_text
+    children_having_extracted_text.docs.map(&:extracted_text).flatten
+  end
+
 end

data/lib/ddr/index/fields.rb

@@ -8,8 +8,11 @@ module Ddr::Index
     ACTIVE_FEDORA_MODEL         = Field.new :active_fedora_model, :stored_sortable
     ADMIN_SET                   = Field.new :admin_set, :stored_sortable
     ADMIN_SET_FACET             = Field.new :admin_set_facet, :facetable
+    ALL_TEXT                    = Field.new :all_text, solr_name: "all_text_timv"
     ASPACE_ID                   = Field.new :aspace_id, :stored_sortable
     ATTACHED_FILES              = Field.new :attached_files, solr_name: "attached_files_ss"
+    ATTACHED_FILES_HAVING_CONTENT =
+      Field.new :attached_files_having_content, :symbol
     BOX_NUMBER_FACET            = Field.new :box_number_facet, :facetable
     COLLECTION_FACET            = Field.new :collection_facet, :facetable
     COLLECTION_URI              = Field.new :collection_uri, :symbol
@@ -24,7 +27,7 @@ module Ddr::Index
     DISPLAY_FORMAT              = Field.new :display_format, :stored_sortable
     DOI                         = Field.new :doi, :symbol
     EAD_ID                      = Field.new :ead_id, :stored_sortable
-    EXTRACTED_TEXT              = Field.new :extracted_text, :searchable, type: :text
+    EXTRACTED_TEXT              = Field.new :extracted_text, solr_name: "extracted_text_tsm"
     FCREPO3_PID                 = Field.new :fcrepo3_pid, :stored_sortable
     HAS_MODEL                   = Field.new :has_model, :symbol
     IDENTIFIER_ALL              = Field.new :identifier_all, :symbol

data/lib/ddr/index/filter.rb

@@ -76,6 +76,10 @@ module Ddr::Index
         term is_member_of_collection: get_id(object_or_id)
       end
 
+      def is_part_of(object_or_id)
+        term is_part_of: get_id(object_or_id)
+      end
+
       def model(*models)
         where active_fedora_model: models
       end

data/lib/ddr/index/query_result.rb

@@ -24,7 +24,7 @@ module Ddr::Index
     def pids
       Enumerator.new do |e|
         each do |doc|
-          e << doc[Fields::PID]
+          e << doc[Fields::ID]
         end
       end
     end

data/lib/ddr/models/base.rb

@@ -43,10 +43,11 @@ module Ddr::Models
       permanent_id && permanent_id.sub(/\Aark:\/\d+\//, "")
     end
 
+    # Cf. https://github.com/duke-libraries/ddr-models/issues/586
     # @see ActiveModel::Conversion
-    def to_key
-      (key = permanent_id_suffix) ? [key] : super
-    end
+    # def to_key
+    #   (key = permanent_id_suffix) ? [key] : super
+    # end
 
     def model_and_id
       "#{self.class} id: #{id.inspect || '[NEW]'}"
@@ -84,7 +85,7 @@ module Ddr::Models
               when :present
                 desc_metadata_terms.select { |t| desc_metadata.values(t).present? }
               when :defined_attributes
-                desc_metadata_terms & desc_metadata_attributes
+                desc_metadata_terms & MetadataMapping.dc11.unqualified_names
               when :required
                 desc_metadata_terms(:defined_attributes).select {|t| required? t}
               when :dcterms
@@ -103,12 +104,6 @@ module Ddr::Models
         terms | desc_metadata_terms(*args)
       end
     end
-    deprecation_deprecate :desc_metadata_terms
-
-    def desc_metadata_attributes
-      MetadataMapping.dc11.unqualified_names
-    end
-    deprecation_deprecate :desc_metadata_attributes
 
     def desc_metadata_values(term)
       Deprecation.warn(Base, "`desc_metadata_values` is deprecated; use `desc_metadata.values` instead.")
@@ -147,7 +142,7 @@ module Ddr::Models
     def attached_files_having_content
       Hash.new.tap do |h|
         attached_files.each do |file_id, file|
-          h[file_id] = file if file.has_content?
+          h[file_id] = file if !file.destroyed? && file.has_content?
         end
       end
     end

data/lib/ddr/models/indexing.rb

@@ -25,6 +25,7 @@ module Ddr
           CONTRIBUTOR_FACET     => desc_metadata.values('contributor'),
           CREATOR_FACET         => desc_metadata.creator,
           DATE_FACET            => desc_metadata.date,
+          ATTACHED_FILES_HAVING_CONTENT => attached_files_having_content.keys,
           DATE_SORT             => date_sort,
           DEPOSITOR             => depositor,
           DISPLAY_FORMAT        => display_format,
@@ -86,6 +87,7 @@ module Ddr
         if is_a? Item
           fields[ADMIN_SET_FACET] = admin_set_facet
           fields[COLLECTION_FACET] = collection_facet
+          fields[ALL_TEXT] = all_text
         end
         fields
       end

data/lib/ddr/models/solr_document.rb

@@ -142,6 +142,10 @@ module Ddr::Models
       has_datastream?(Ddr::Models::File::CONTENT)
     end
 
+    def has_extracted_text?
+      has_datastream?(Ddr::Datastreams::EXTRACTED_TEXT)
+    end
+
     def content_ds
       datastreams[Ddr::Models::File::CONTENT]
     end
@@ -228,6 +232,10 @@ module Ddr::Models
       end
     end
 
+    def published?
+      self[Ddr::Index::Fields::WORKFLOW_STATE] == Ddr::Managers::WorkflowManager::PUBLISHED
+    end
+
     private
 
     def targets_query

data/lib/ddr/models/version.rb

@@ -1,5 +1,5 @@
 module Ddr
   module Models
-    VERSION = "3.0.0.beta.17"
+    VERSION = "3.0.0.beta.18"
   end
 end

data/spec/auth/ability_spec.rb

@@ -32,7 +32,8 @@ module Ddr::Auth
       end
 
       describe "non-downloadable attached_files" do
-        (Component.ds_specs.keys.map(&:to_s) - DatastreamAbilityDefinitions::DATASTREAM_DOWNLOAD_ABILITIES.keys).each do |dsid|
+        (Component.child_resource_reflections.keys.map(&:to_s) - \
+                          DatastreamAbilityDefinitions::DATASTREAM_DOWNLOAD_ABILITIES.keys).each do |dsid|
           describe "\"#{dsid}\"" do
             let(:ds) { obj.attached_files[dsid] }
             before { subject.can :download, obj.id }

data/spec/fixtures/extractedText1.txt

@@ -0,0 +1 @@
+Lorem ipsum dolor sit amet, consectetur adipiscing elit. Duis scelerisque diam in lacus sollicitudin maximus. Duis molestie rutrum hendrerit. Vestibulum lobortis augue a ultrices mollis. Mauris semper eu odio posuere scelerisque. Donec vel condimentum nunc. Cum sociis natoque penatibus et magnis dis parturient montes, nascetur ridiculus mus. Morbi fermentum, sem vel interdum suscipit, lacus risus ultricies sem, in lacinia ligula risus ut est. Integer gravida, orci ac lobortis blandit, eros tortor bibendum turpis, quis semper velit erat in augue.

data/spec/fixtures/extractedText2.txt

@@ -0,0 +1 @@
+Proin convallis nibh quis ex iaculis, eget porttitor elit consequat. Duis in tincidunt ipsum. Aenean imperdiet lacinia libero nec egestas. Fusce a metus volutpat, vehicula augue accumsan, mattis lorem. Donec ex tellus, tempus vitae orci sit amet, sollicitudin pharetra velit. Fusce egestas nunc vitae placerat elementum. Nullam pharetra porta eros, quis tempor sem pulvinar a. Ut in neque cursus, hendrerit neque in, congue est. Phasellus neque mauris, eleifend in sem ac, posuere tincidunt odio. Duis lectus nibh, ultricies a elementum a, consectetur vitae turpis.

data/spec/fixtures/extractedText3.txt

@@ -0,0 +1,3 @@
+Ut eu consequat turpis. Proin commodo facilisis pretium. Vestibulum tristique volutpat lobortis. Donec efficitur tincidunt porta. Sed dui massa, rutrum iaculis interdum at, scelerisque nec velit. Suspendisse potenti. Aliquam quis mauris sed magna convallis ullamcorper non ut ipsum.
+
+Donec dolor urna, malesuada id condimentum ut, feugiat consectetur lectus. Morbi dignissim nibh neque, a bibendum ex semper vel. Nunc pellentesque varius bibendum. Etiam posuere lorem nibh, sit amet iaculis dolor lobortis nec. Nam congue dui non lectus luctus scelerisque. Aliquam turpis orci, porta non dapibus eu, accumsan sed tortor. Phasellus placerat ac mauris sed vehicula. Sed dolor massa, fringilla vitae urna vel, efficitur porta eros. Ut massa nisi, bibendum id mollis quis, aliquet eu purus. Aliquam erat volutpat. Phasellus accumsan felis eget dui vulputate, eget feugiat diam eleifend. Donec facilisis massa odio, ac auctor eros ullamcorper quis. Donec sit amet malesuada lectus. Aenean egestas, leo id lacinia mattis, dui ex tempor tellus, vel mollis velit leo a ante.

data/spec/models/component_spec.rb

@@ -15,4 +15,68 @@ RSpec.describe Component, type: :model, components: true do
     its(:index_fields) { is_expected.to include(Ddr::Index::Fields::COLLECTION_URI => "test-1") }
   end
 
+  describe "extracted text" do
+    let(:parent) { FactoryGirl.create(:item) }
+    describe "when the child is not already associated with the parent" do
+      before {
+        subject.extractedText.content = fixture_file_upload('extractedText1.txt')
+        subject.save
+        subject.reload
+      }
+      it "updates the parent index when associated" do
+        expect(subject).to receive(:index_parent)
+        subject.parent = parent
+        subject.save
+      end
+    end
+    describe "when the child is already associated with the parent" do
+      before {
+        subject.parent = parent
+        subject.save
+      }
+      describe "when there is no extracted text" do
+        describe "and none is added" do
+          it "does not trigger an index update on the parent" do
+            expect(subject).not_to receive(:index_parent)
+            subject.save
+          end
+        end
+        describe "and extracted text is added" do
+          it "triggers an index update on the parent" do
+            expect(subject).to receive(:index_parent)
+            subject.extractedText.content = fixture_file_upload('extractedText1.txt')
+            subject.save
+          end
+        end
+      end
+      describe "when extracted text exists" do
+        before {
+          subject.extractedText.content = fixture_file_upload('extractedText1.txt')
+          subject.save
+          subject.reload
+        }
+        describe "and is removed" do
+          it "triggers an index update on the parent" do
+            pending "Deleting a datastream does not mark the content as changed"
+            expect(subject).to receive(:index_parent)
+            subject.extractedText.delete
+            subject.save
+          end
+        end
+        describe "and doesn't change" do
+          it "triggers an index update on the parent" do
+            expect(subject).to receive(:index_parent)
+            subject.save
+          end
+        end
+        describe "and changes" do
+          it "triggers an index update on the parent" do
+            expect(subject).to receive(:index_parent)
+            subject.extractedText.content = fixture_file_upload('extractedText2.txt')
+            subject.save
+          end
+        end
+      end
+    end
+  end
 end

data/spec/models/indexing_spec.rb

@@ -57,6 +57,9 @@ module Ddr::Models
       }
 
       its([Indexing::CONTENT_CREATE_DATE]) { is_expected.to eq "2016-01-22T21:50:33Z" }
+      its([Indexing::ATTACHED_FILES_HAVING_CONTENT]) {
+        is_expected.to eq([:content])
+      }
     end
 
   end

data/spec/models/item_spec.rb

@@ -6,4 +6,27 @@ RSpec.describe Item, type: :model do
   it_behaves_like "a non-collection model"
   it_behaves_like "a potentially publishable object"
 
+  describe "indexing text" do
+    let(:children) { FactoryGirl.build_list(:component, 5) }
+
+    let(:text1) { fixture_file_upload('extractedText1.txt', 'text/plain') }
+    let(:text2) { fixture_file_upload('extractedText2.txt', 'text/plain') }
+    let(:text3) { fixture_file_upload('extractedText3.txt', 'text/plain') }
+
+    before {
+      children[0].extractedText.content = text1
+      children[0].save
+      children[1].extractedText.content = text2
+      children[1].save
+      children[2].extractedText.content = text3
+      children[2].save
+      subject.children = children
+      subject.save
+    }
+
+    it "indexes the combined text of its children" do
+      expect(subject.index_fields[Ddr::Index::Fields::ALL_TEXT]).to contain_exactly(text1.read, text2.read, text3.read)
+    end
+  end
+
 end

data/spec/models/solr_document_spec.rb

@@ -212,4 +212,14 @@ RSpec.describe SolrDocument, type: :model, contacts: true do
     end
   end
 
+  describe "#published?" do
+    context "when the object is published" do
+      before { subject[Ddr::Index::Fields::WORKFLOW_STATE] = Ddr::Managers::WorkflowManager::PUBLISHED }
+      its(:published?) { is_expected.to eq(true) }
+    end
+    context "when the object is published" do
+      its(:published?) { is_expected.to eq(false) }
+    end
+  end
+
 end

data/spec/support/shared_examples_for_ddr_models.rb

@@ -14,7 +14,10 @@ RSpec.shared_examples "a DDR model" do
       before {
         subject.permanent_id = "ark:/99999/fk4rx95k8w"
       }
-      its(:to_param) { is_expected.to eq("fk4rx95k8w") }
+      its(:to_param) do
+        pending "Resolution of https://github.com/duke-libraries/ddr-models/issues/586"
+        is_expected.to eq("fk4rx95k8w")
+      end
     end
     describe "when it does not have a permanent id" do
       its(:to_param) { is_expected.to_not be_nil }

data/spec/support/shared_examples_for_has_content.rb

@@ -23,14 +23,12 @@ RSpec.shared_examples "an object that can have content" do
   describe "extracted text" do
     describe "when it is not present" do
       its(:has_extracted_text?) { should be false }
-      its(:to_solr) { should_not include(Ddr::Index::Fields::EXTRACTED_TEXT) }
+      its(:to_solr) { is_expected.not_to include(Ddr::Index::Fields::EXTRACTED_TEXT) }
     end
     describe "when it is present" do
       before { subject.extractedText.content = "This is my text. See Spot run." }
       its(:has_extracted_text?) { should be true }
-      it "should be indexed" do
-        expect(subject.to_solr[Ddr::Index::Fields::EXTRACTED_TEXT]).to eq("This is my text. See Spot run.")
-      end
+      its(:to_solr) { is_expected.to include(Ddr::Index::Fields::EXTRACTED_TEXT) }
     end
   end
 

metadata

@@ -1,7 +1,7 @@
 --- !ruby/object:Gem::Specification
 name: ddr-models
 version: !ruby/object:Gem::Version
-  version: 3.0.0.beta.17
+  version: 3.0.0.beta.18
 platform: ruby
 authors:
 - Jim Coble
@@ -9,7 +9,7 @@ authors:
 autorequire: 
 bindir: bin
 cert_chain: []
-date: 2016-03-09 00:00:00.000000000 Z
+date: 2016-03-11 00:00:00.000000000 Z
 dependencies:
 - !ruby/object:Gem::Dependency
   name: rails
@@ -662,6 +662,9 @@ files:
 - spec/fixtures/8bit.tif
 - spec/fixtures/arrow1rightred_e0.gif
 - spec/fixtures/bird.jpg
+- spec/fixtures/extractedText1.txt
+- spec/fixtures/extractedText2.txt
+- spec/fixtures/extractedText3.txt
 - spec/fixtures/fits/document.xml
 - spec/fixtures/fits/image.xml
 - spec/fixtures/imageA.tif
@@ -837,6 +840,9 @@ test_files:
 - spec/fixtures/8bit.tif
 - spec/fixtures/arrow1rightred_e0.gif
 - spec/fixtures/bird.jpg
+- spec/fixtures/extractedText1.txt
+- spec/fixtures/extractedText2.txt
+- spec/fixtures/extractedText3.txt
 - spec/fixtures/fits/document.xml
 - spec/fixtures/fits/image.xml
 - spec/fixtures/imageA.tif