fedora-migrate 0.0.1 → 0.1.0

data/spec/unit/content_mover_spec.rb

@@ -0,0 +1,78 @@
+require 'spec_helper'
+
+describe FedoraMigrate::ContentMover do
+
+  let(:nil_source) { double("Source", content: nil, dsid: "datastream") }
+  let(:source) do
+    double("Source", 
+      content: "foo", 
+      dsid: "datastream",
+      label: "label",
+      mimeType: "mimetype",
+      createDate: Time.new(1993, 02, 24, 12, 0, 0, "+09:00")  # Rubydora returns Time objects for datastreams' creation dates
+    )
+  end
+  let(:target) { double("Target", content: "") }
+
+  describe "#migrate" do
+    context "without content" do
+      subject { FedoraMigrate::ContentMover.new(nil_source, target).migrate }
+      it { is_expected.to be true }
+    end
+    context "with content" do
+      subject { FedoraMigrate::ContentMover.new(source, target).migrate }
+      before do
+        allow_any_instance_of(FedoraMigrate::ContentMover).to receive(:move_content).and_return(true)
+        allow_any_instance_of(FedoraMigrate::ContentMover).to receive(:insert_date_created_by_application).and_return(true)
+      end
+      it { is_expected.to be true }
+    end
+  end
+
+  describe "#move_content" do
+    before do
+      allow(target).to receive(:content=).with("foo")
+      allow(target).to receive(:original_name=).with("label")
+      allow(target).to receive(:mime_type=).with("mimetype")
+      allow(target).to receive(:save).and_return(true)
+      allow_any_instance_of(FedoraMigrate::ContentMover).to receive(:insert_date_created_by_application).and_return(true)
+    end
+    subject do  
+      FedoraMigrate::ContentMover.new(source, target).move_content
+    end
+    it { is_expected.to be true }
+  end
+
+  describe "#insert_date_created_by_application" do
+    subject { FedoraMigrate::ContentMover.new(source, target).insert_date_created_by_application }
+    context "with a successful update" do
+      let(:successful_status) { double("Result", status: 204) }
+      before { allow_any_instance_of(FedoraMigrate::ContentMover).to receive(:perform_sparql_insert).and_return(successful_status) }
+      it { is_expected.to be true }
+    end
+    context "with an unsuccessful update" do
+      let(:unsuccessful_status) { double("Result", status: 404, body: "Error!") }
+      before { allow_any_instance_of(FedoraMigrate::ContentMover).to receive(:perform_sparql_insert).and_return(unsuccessful_status) }
+      it "should raise an error" do
+        expect { subject }.to raise_error FedoraMigrate::Errors::MigrationError
+      end
+    end
+  end
+
+  describe "#sparql_insert" do
+    let(:sample_sparql_query) do
+<<-EOF
+PREFIX premis: <http://www.loc.gov/premis/rdf/v1#>
+PREFIX xsd: <http://www.w3.org/2001/XMLSchema#>
+DELETE WHERE { ?s premis:hasDateCreatedByApplication ?o } ;
+INSERT {
+  <> premis:hasDateCreatedByApplication "1993-02-24T12:00:00+09:00"^^xsd:dateTime .
+}
+WHERE { }
+EOF
+    end
+    subject { FedoraMigrate::ContentMover.new(source, target).sparql_insert }
+    it { is_expected.to eql sample_sparql_query }
+  end
+
+end

data/spec/unit/datastream_verification_spec.rb

@@ -0,0 +1,60 @@
+require 'spec_helper'
+
+describe FedoraMigrate::DatastreamVerification do
+
+  class TestSubject
+    include FedoraMigrate::DatastreamVerification
+    def initialize datastream
+      @datastream = datastream
+      @source = datastream
+    end
+  end
+
+  describe "binary sources from Fedora3" do
+    let(:bad_binary_source)     { double("Datastream", checksum: "bad",     mimeType: "binary", content: "XXXXXX", dsid: "content", pid: "abc123") }
+    let(:good_binary_source)    { double("Datastream", checksum: "foo",     mimeType: "binary", content: "foo",    dsid: "content", pid: "abc123") }
+    let(:missing_checksum)      { double("Datastream", checksum: "missing", mimeType: "binary", content: "foo",    dsid: "content", pid: "abc123") }
+    context "that match Fedora4's checksum" do
+      subject { TestSubject.new(good_binary_source) }
+      before  { allow(subject).to receive(:target_checksum).once.and_return("foo") }
+      it      { is_expected.to have_matching_checksums }
+      it      { is_expected.to be_valid }
+    end
+    context "that do not match Fedora4's checksum" do
+      subject { TestSubject.new(bad_binary_source) }
+      before  { allow(subject).to receive(:target_checksum).twice.and_return("bar") }
+      specify "are not valid and logged" do
+        expect(FedoraMigrate::Logger).to receive(:warn)
+        expect(subject).to_not be_valid
+      end
+    end
+    context "when the checksum is missing" do
+      subject { TestSubject.new(missing_checksum) }
+      context "and a newly calculated checksum matches" do
+        before { allow(subject).to receive(:target_checksum).twice.and_return(Digest::SHA1.hexdigest("foo")) }
+        it     { is_expected.to have_matching_checksums }
+        it     { is_expected.to be_valid }
+      end
+      context "and a newly calculated checksum does not match" do
+        before { expect_any_instance_of(TestSubject).to receive(:target_checksum).twice.and_return(Digest::SHA1.hexdigest("bar")) }
+        specify "are not valid and logged" do
+          expect(FedoraMigrate::Logger).to receive(:warn)
+          expect(subject).to_not be_valid
+        end
+      end
+    end
+  end
+
+  describe "xml sources from Fedora3" do
+    subject { TestSubject.new(double("Datastream", checksum: "invalid", mimeType: "text/xml", content: "<bar></bar>")) }
+    context "when the datastream content is correctly altered upon migration" do
+      before  { allow(subject).to receive(:target_content).once.and_return("<?xml version=\"1.0\"?>\n<bar></bar>") }
+      it      { is_expected.to have_matching_nokogiri_checksums }
+    end
+    context "when the datastream content is incorrectly altered upon migration" do
+      before { allow(subject).to receive(:target_content).once.and_return("<?xml version=\"1.0\"?>\n<baz></baz>") }
+      it     { is_expected.to_not have_matching_nokogiri_checksums }
+    end
+  end
+
+end

data/spec/unit/dates_mover_spec.rb

@@ -0,0 +1,33 @@
+require 'spec_helper'
+
+describe FedoraMigrate::DatesMover do
+  let(:target) { ExampleModel::RDFObject.new }
+  let(:source) { instance_double('Source', createdDate: 'yesterday', lastModifiedDate: 'today') }
+
+  subject { FedoraMigrate::DatesMover.new(source, target) }
+
+  describe '#migrate' do
+
+    it 'migrates the create and mod dates' do
+      subject.migrate
+      expect(target.date_uploaded).to eq 'yesterday'
+      expect(target.date_modified).to eq 'today'
+    end
+
+    context "when the source methods don't exist" do
+      let(:source) { instance_double('Source with no date methods') }
+      it 'gracefully does nothing' do
+        expect{ subject.migrate }.to_not raise_error
+        expect(target.date_uploaded).to be_nil
+        expect(target.date_modified).to be_nil
+      end
+    end
+
+    context "when the target methods don't exist" do
+      let(:target) { instance_double('Target with no date methods') }
+      it 'gracefully does nothing' do
+        expect{ subject.migrate }.to_not raise_error
+      end
+    end
+  end
+end

data/spec/unit/migration_options_spec.rb

@@ -0,0 +1,61 @@
+require 'spec_helper'
+
+describe FedoraMigrate::MigrationOptions do
+
+  class TestCase
+    include FedoraMigrate::MigrationOptions
+  end
+
+  describe "#conversion_options" do
+    subject do
+      TestCase.new.tap do |example|
+        example.options = { convert: 'datastream' }
+      end
+    end
+    specify "sets the name of the datastream to convert" do
+      expect(subject.conversion_options).to include "datastream"
+    end
+    it { is_expected.to be_not_forced }
+  end
+
+  describe "#forced?" do
+    context "when set to true" do
+      subject do
+        TestCase.new.tap do |example|
+          example.options = { convert: "datastream", force: true }
+        end
+      end
+      it { is_expected.to be_forced }
+    end
+    context "when set to false" do
+      subject do
+        TestCase.new.tap do |example|
+          example.options = { force: false }
+        end
+      end
+      it { is_expected.to be_not_forced }
+    end
+    context "by default" do
+      subject { TestCase.new }
+      it { is_expected.to be_not_forced }
+    end
+  end
+
+  describe "#application_creates_versions" do
+    context "by default" do
+      subject do
+        TestCase.new.application_creates_versions?
+      end
+      it { is_expected.to be false }
+    end
+    context "when our own Hydra application creates versions" do
+      subject do
+        TestCase.new.tap do |example|
+          example.options = { application_creates_versions: true }
+        end
+      end
+      it { is_expected.to be_application_creates_versions }
+    end
+  end
+
+end

data/spec/unit/mover_spec.rb

@@ -7,7 +7,6 @@ describe FedoraMigrate::Mover do
   it { is_expected.to respond_to :options }
 
   describe "#new" do
-    
     context "with two arguments" do
       subject { FedoraMigrate::Mover.new("foo", "bar") }
       specify "has a source" do
@@ -34,6 +33,41 @@ describe FedoraMigrate::Mover do
         expect(subject.target).to be_nil
       end
     end
+  end
 
+  describe "::id_component" do
+    context "with a Rubydora object" do
+      let(:id)      { "rb68xc11m" }
+      let(:object)  { FedoraMigrate.source.connection.find("sufia:#{id}") }
+      subject { FedoraMigrate::Mover.id_component(object) }
+      it { is_expected.to eql(id) }
+    end
+    context "with a URI" do
+      let(:object)  { RDF::URI.new("foo:bar") }
+      subject { FedoraMigrate::Mover.id_component(object) }
+      it { is_expected.to eql("bar") }
+    end
+    context "with a string" do
+      let(:object)  { "foo:bar" }
+      subject { FedoraMigrate::Mover.id_component(object) }
+      it { is_expected.to eql("bar") }
+    end
   end
+
+  describe "#id_component" do
+    context "with a source" do
+      subject { FedoraMigrate::Mover.new("source:pid").id_component }
+      it { is_expected.to eql("pid") }
+    end
+    context "object, but no source" do
+      subject { FedoraMigrate::Mover.new.id_component("source:pid") }
+      it { is_expected.to eql("pid") }
+    end
+    context "neither object, nor source" do
+      specify "raises an error" do
+        expect { FedoraMigrate::Mover.new.id_component }.to raise_error(FedoraMigrate::Errors::MigrationError)
+      end
+    end
+  end
+
 end

data/spec/unit/object_mover_spec.rb

@@ -7,7 +7,6 @@ describe FedoraMigrate::ObjectMover do
   end
 
   describe "#new" do
-
     it { is_expected.to respond_to :source }
     it { is_expected.to respond_to :target }
     it { is_expected.to respond_to :post_initialize }
@@ -19,8 +18,7 @@ describe FedoraMigrate::ObjectMover do
     end
     it "should call the before hook and save the target" do
       expect_any_instance_of(FedoraMigrate::ObjectMover).to receive(:before_object_migration)
-      expect_any_instance_of(FedoraMigrate::ObjectMover).to receive(:save).and_return(true)
-      expect(subject).to be true
+      expect(subject).to be nil
     end
   end
 

data/spec/unit/rels_ext_datastream_mover_spec.rb

@@ -2,35 +2,45 @@ require 'spec_helper'
 
 describe FedoraMigrate::RelsExtDatastreamMover do
 
-  context "with a target" do
+  let(:file_id)  { "rb68xc11m" }
+  let(:batch_id) { "rb68xc09k" }
+  let(:source)   { FedoraMigrate.source.connection.find("sufia:#{file_id}") }
+  let(:query)    { subject.target }
 
+  context "with target objects present in Fedora 4" do
     before do
-      allow_any_instance_of(FedoraMigrate::RelsExtDatastreamMover).to receive(:retrieve_subject).and_return("subject")
-      allow_any_instance_of(FedoraMigrate::RelsExtDatastreamMover).to receive(:retrieve_object).and_return("object")
+      ActiveFedora::Base.create(id: file_id)
+      ActiveFedora::Base.create(id: batch_id)
     end
 
-    subject do
-      FedoraMigrate::RelsExtDatastreamMover.new(
-        FedoraMigrate.source.connection.find("sufia:rb68xc11m")
-      )
-    end
-
-    describe "#relationships" do
-      it "should parse the source's RELS-EXT datastream for relationships" do
-        expect(subject.relationships).to include(:part_of => ["object"])
+    describe "#initialize" do
+      context "without a target" do
+        subject { FedoraMigrate::RelsExtDatastreamMover.new(source).target }
+        it { is_expected.to be_kind_of(ActiveFedora::Base) }
+      end
+      context "with a supplied target" do
+        subject { FedoraMigrate::RelsExtDatastreamMover.new(source, "a target").target }
+        it { is_expected.to eql "a target" }
       end
     end
 
-    describe "#ng_xml" do
-      it "should return a Nokogiri document of the object's RELS-EXT datastream" do
-        expect(subject.ng_xml).to be_kind_of(Nokogiri::XML::Document)
+    describe "#migrate" do
+      context "with an existing target" do
+        before  { FedoraMigrate::RelsExtDatastreamMover.new(source).migrate }
+        subject { ActiveFedora::Base.find(file_id).ldp_source.graph.query([nil, ActiveFedora::RDF::Fcrepo::RelsExt.isPartOf, nil]) }
+        it "migrates RDF relationships" do
+          expect(subject.first.subject).to eq RDF::URI.new("http://localhost:8983/fedora/rest/test/#{file_id}")
+          expect(subject.first.object).to eq RDF::URI.new("http://localhost:8983/fedora/rest/test/#{batch_id}")
+        end
       end
     end
+  end
 
-    describe "#has_relationships?" do
-      it { is_expected.to have_relationships }
+  context "with a non-existent target" do
+    let(:error_message) { "Target object was not found in Fedora 4. Did you migrate it?" }
+    it "raises an error" do
+      expect { FedoraMigrate::RelsExtDatastreamMover.new(source) }.to raise_error(FedoraMigrate::Errors::MigrationError, error_message)
     end
-
   end
 
 end

data/spec/unit/repository_migrator_spec.rb

@@ -5,12 +5,23 @@ describe FedoraMigrate::RepositoryMigrator do
   let(:namespace) { "sufia" }
 
   it { is_expected.to respond_to(:source_objects) }
-  it { is_expected.to respond_to(:results) }
+  it { is_expected.to respond_to(:failed) }
   it { is_expected.to respond_to(:namespace) }
 
-  describe "#results" do
-    specify "are initially empty" do
-      expect(subject.results).to eql([])
+  describe "#failed" do
+    specify "sets to zero" do
+      expect(subject.failed).to eql(0)
+    end
+  end
+
+  context "when forcing" do
+    before do
+      allow_any_instance_of(FedoraMigrate::RepositoryMigrator).to receive(:source_objects).and_return([])
+      allow_any_instance_of(FedoraMigrate::RepositoryMigrator).to receive(:failed).and_return(1) 
+    end
+    subject { FedoraMigrate::RepositoryMigrator.new(namespace, { force: true }) }
+    specify "migrate relationships if failures are present" do
+      expect(subject.migrate_relationships).to be true
     end
   end
 
@@ -35,7 +46,7 @@ describe FedoraMigrate::RepositoryMigrator do
       end
       it "should exclude fedora-system objects" do
         expect(subject.source_objects).to_not include("fedora-system:ContentModel-3.0")
-        expect(subject.source_objects.count).to eql 5
+        expect(subject.source_objects.count).to eql 9
       end
     end
   end

data/spec/unit/target_constructor_spec.rb

@@ -0,0 +1,34 @@
+require 'spec_helper'
+
+describe FedoraMigrate::TargetConstructor do
+
+  context "with one qualified model" do
+    let(:list) { ["info:fedora/fedora-system:FedoraObject-3.0", "info:fedora/afmodel:String"] }
+    subject { FedoraMigrate::TargetConstructor.new(list).build }
+    it "should choose the one that is valid" do
+      expect(subject.target).to eql String
+    end
+  end
+
+  context "with multiple qualified models" do
+    let(:list) { ["info:fedora/fedora-system:FedoraObject-3.0", "info:fedora/afmodel:Array", "info:fedora/afmodel:String"] }
+    subject { FedoraMigrate::TargetConstructor.new(list).build }
+    it "should choose the first one that is valid" do
+      expect(subject.target).to eql Array
+    end
+  end
+
+  context "with a single qualified model" do
+    subject { FedoraMigrate::TargetConstructor.new("info:fedora/afmodel:Array").build }
+    it "should be valid" do
+      expect(subject.target).to eql Array
+    end
+  end
+
+  context "with multiple unqualified models" do
+    let(:list) { ["info:fedora/fedora-system:FedoraObject-3.0", "info:fedora/fedora-system:FooObject"] }
+    subject { FedoraMigrate::TargetConstructor.new(list).build.target }
+    it { is_expected.to be_nil }
+  end
+
+end

data/tasks/dev.rake

@@ -14,7 +14,7 @@ namespace :fixtures do
   desc "Load Fedora3 fixtures for testing; use FIXTURE_PATH= for your own"
   task :load do
     repo = FedoraMigrate.source
-    path = ENV["FIXTURE_PATH"] || "spec/fixtures/objects" 
+    path = ENV["FIXTURE_PATH"] || "spec/fixtures/objects"
     Dir.glob(File.join(path,"*.xml")).each do |f|
       fixture = File.open(f)
       begin