fedora-migrate 0.1.0 → 0.2.0

data/spec/fixtures/sample-report.json

@@ -0,0 +1,166 @@
+{
+  "scholarsphere:000000000": {
+    "status": true,
+    "object": {
+      "id": "000000000",
+      "class": "Batch",
+      "content_datastreams": [
+
+      ],
+      "rdf_datastreams": [
+        {
+          "ds": "descMetadata",
+          "status": [
+
+          ]
+        }
+      ],
+      "permissions": null,
+      "dates": {
+        "uploaded": null,
+        "modified": null
+      }
+    },
+    "relationships": [
+
+    ]
+  },
+  "scholarsphere:000000018": {
+    "status": true,
+    "object": {
+      "id": "000000018",
+      "class": "GenericFile",
+      "content_datastreams": [
+        {
+          "ds": "characterization",
+          "versions": [
+            {
+              "name": "",
+              "mime_type": "text/xml",
+              "original_date": "2013-03-15T11:08:21Z"
+            }
+          ]
+        },
+        {
+          "ds": "content",
+          "versions": [
+            {
+              "name": "Open_Up_Your_RepositoryWith_a_SWORD_.pdf",
+              "mime_type": "application/pdf",
+              "original_date": "2012-09-26T16:09:14Z"
+            },
+            {
+              "name": "Open_Up_Your_RepositoryWith_a_SWORD_.pdf",
+              "mime_type": "application/pdf",
+              "original_date": "2012-11-26T03:11:55Z"
+            },
+            {
+              "name": "Open_Up_Your_RepositoryWith_a_SWORD_.pdf",
+              "mime_type": "application/pdf",
+              "original_date": "2012-11-26T03:12:07Z"
+            },
+            {
+              "name": "Open_Up_Your_RepositoryWith_a_SWORD_.pdf",
+              "mime_type": "application/pdf",
+              "original_date": "2012-11-26T03:12:10Z"
+            },
+            {
+              "name": "Open_Up_Your_RepositoryWith_a_SWORD_.pdf",
+              "mime_type": "application/pdf",
+              "original_date": "2012-11-26T03:12:13Z"
+            },
+            {
+              "name": "Open_Up_Your_RepositoryWith_a_SWORD_.pdf",
+              "mime_type": "application/pdf",
+              "original_date": "2012-11-26T03:12:17Z"
+            },
+            {
+              "name": "Open_Up_Your_RepositoryWith_a_SWORD_.pdf",
+              "mime_type": "application/pdf",
+              "original_date": "2012-11-26T03:13:12Z"
+            }
+          ]
+        },
+        {
+          "ds": "thumbnail",
+          "versions": [
+            {
+              "name": "",
+              "mime_type": "image/jpeg",
+              "original_date": "2014-09-10T15:12:43Z"
+            }
+          ]
+        },
+        {
+          "ds": "full_text",
+          "versions": [
+            {
+              "name": "File Datastream",
+              "mime_type": "application/octet-stream",
+              "original_date": "2013-03-14T10:03:13Z"
+            }
+          ]
+        }
+      ],
+      "rdf_datastreams": [
+        {
+          "ds": "descMetadata",
+          "status": [
+
+          ]
+        }
+      ],
+      "permissions": [
+        "read_groups = [\"public\"]",
+        "edit_groups = []",
+        "discover_groups = []",
+        "read_users = []",
+        "edit_users = [\"mjg36\"]",
+        "discover_users = []"
+      ],
+      "dates": {
+        "uploaded": "2012-09-26T16:09:13.394Z",
+        "modified": "2015-01-02T22:16:02.961Z"
+      }
+    },
+    "relationships": [
+      "http://ssrepo2qa.dlt.psu.edu:8080/SSqaFedora4/rest/prod/00/00/00/01/000000018--info:fedora/fedora-system:def/relations-external#isPartOf--http://ssrepo2qa.dlt.psu.edu:8080/SSqaFedora4/rest/prod/00/00/00/00/000000000"
+    ]
+  },
+  "scholarsphere:05741r698": {
+    "status": true,
+    "object": {
+      "id": "05741r698",
+      "class": "Batch",
+      "content_datastreams": [
+
+      ],
+      "rdf_datastreams": [
+        {
+          "ds": "descMetadata",
+          "status": [
+
+          ]
+        }
+      ],
+      "permissions": null,
+      "dates": {
+        "uploaded": null,
+        "modified": null
+      }
+    },
+    "relationships": [
+
+    ]
+  },
+  "scholarsphere:6395wb555": {
+    "status": false,
+    "object": "#<Ldp::BadRequest: RDF was not parsable>",
+    "relationships": "#<FedoraMigrate::Errors::MigrationError: Target object was not found in Fedora 4. Did you migrate it?>"
+  },
+  "scholarsphere:x346dm27k": {
+    "status": false,
+    "object": "#<Ldp::BadRequest: RDF was not parsable>",
+    "relationships": "#<FedoraMigrate::Errors::MigrationError: Target object was not found in Fedora 4. Did you migrate it?>"
+  }
+}

data/spec/integration/missing_relationships_spec.rb

@@ -12,19 +12,28 @@ describe "Collections with missing files" do
   end
 
   context "when migrating relationships" do
-
     let(:migrated_collection) { ExampleModel::Collection.first }
     let(:error_message) do
       "scholarsphere:#{collection} could not migrate relationship info:fedora/fedora-system:def/relations-external#hasCollectionMember because info:fedora/scholarsphere:#{missing_file} doesn't exist in Fedora 4"
-    end 
-
-    it "should only migrate existing relationships and log failed ones" do
-      expect(FedoraMigrate::Logger).to receive(:warn).with(error_message)
-      FedoraMigrate::RelsExtDatastreamMover.new(FedoraMigrate.find("scholarsphere:#{collection}")).migrate
+    end
+    before { FedoraMigrate::RelsExtDatastreamMover.new(FedoraMigrate.find("scholarsphere:#{collection}")).migrate }
+    it "should only migrate existing relationships" do
       expect(migrated_collection.members.count).to eql 2
       expect(migrated_collection.member_ids).to_not include(missing_file)
     end
+  end
 
+  context "when reporting" do
+    subject { FedoraMigrate::RelsExtDatastreamMover.new(FedoraMigrate.find("scholarsphere:#{collection}")).migrate }
+    it "should include failed relationships" do
+      expect(subject.sort.first).to match(/^could not migrate relationship/)
+    end
+    it "should include all the possible relationships" do
+      expect(subject.count).to eql 3
+    end
+    it "should include the successful relationships" do
+      expect(subject.sort.last).to match(/^http/)
+    end
   end
 
 end

data/spec/integration/repository_migration_spec.rb

@@ -10,9 +10,10 @@ describe "Migrating the repository" do
       Object.send(:remove_const, :Collection) if defined?(Collection)
     end
 
-    it "should log warnings" do
-      expect(FedoraMigrate::Logger).to receive(:warn).at_least(6).times
-      FedoraMigrate.migrate_repository(namespace: "sufia", options: {convert: "descMetadata"})
+    subject { FedoraMigrate.migrate_repository(namespace: "sufia", options: {convert: "descMetadata"}).report }
+
+    it "should report warnings" do
+      expect(subject.failed_objects.count).to eql 9
     end
   end
 
@@ -51,9 +52,7 @@ describe "Migrating the repository" do
     let(:versions_report) { GenericFile.all.map { |f| f.content.versions.count }.uniq }
 
     context "by default" do
-      before do
-        FedoraMigrate.migrate_repository(namespace: "sufia", options: {convert: "descMetadata"})
-      end
+      before { FedoraMigrate.migrate_repository(namespace: "sufia", options: {convert: "descMetadata"}) }
       it "should move every object and its versions" do
         expect(file1.title).to eql ["world.png"]
         expect(file2.title).to eql ["Sample Migration Object A"]
@@ -80,7 +79,27 @@ describe "Migrating the repository" do
       end
     end
 
+    context "with an existing report" do
+      let(:report) { "spec/fixtures/failed-report.json" }
+      let(:new_report) { FedoraMigrate::MigrationReport.new("report.json") }
+      before do
+        FileUtils.rm("report.json") if File.exists?("report.json")
+        migrator = FedoraMigrate.migrate_repository(namespace: "sufia", options: {convert: "descMetadata", report: report})
+        migrator.report.save
+      end
+      after { FileUtils.rm("report.json") }
+      it "only migrates the objects that have failed" do
+        expect(GenericFile.all.count).to eql 1
+        expect(Batch.all.count).to eql 1
+        expect(Collection.all.count).to eql 0
+        expect(new_report.total_objects).to eql 9
+        expect(new_report.failures).to eql 0
+      end
+    end
+
   end
 
+
+
 end
 

data/spec/unit/content_mover_spec.rb

@@ -12,12 +12,15 @@ describe FedoraMigrate::ContentMover do
       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: "") }
+  let(:target) { double("Target", content: "", original_name: nil, mime_type: nil) }
 
   describe "#migrate" do
     context "without content" do
       subject { FedoraMigrate::ContentMover.new(nil_source, target).migrate }
-      it { is_expected.to be true }
+      it "reports a nil source" do
+        expect(subject).to be_kind_of FedoraMigrate::ContentMover::Report
+        expect(subject.error).to eql "Nil source -- it's probably defined in the target but not present in the source"
+      end
     end
     context "with content" do
       subject { FedoraMigrate::ContentMover.new(source, target).migrate }
@@ -25,7 +28,7 @@ describe FedoraMigrate::ContentMover 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 }
+      it { is_expected.to be_kind_of FedoraMigrate::ContentMover::Report }
     end
   end
 
@@ -34,13 +37,19 @@ describe FedoraMigrate::ContentMover 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)
+      allow(target).to receive(:save).and_return(true)   
     end
     subject do  
       FedoraMigrate::ContentMover.new(source, target).move_content
     end
-    it { is_expected.to be true }
+    context "with a valid checksum" do
+      before { allow_any_instance_of(FedoraMigrate::ContentMover).to receive(:valid?).and_return(true) }
+      it { is_expected.to be nil }
+    end
+    context "with an invalid checksum" do
+      before { allow_any_instance_of(FedoraMigrate::ContentMover).to receive(:valid?).and_return(false) }
+      it { is_expected.to eql "Failed checksum" }
+    end
   end
 
   describe "#insert_date_created_by_application" do
@@ -48,14 +57,12 @@ describe FedoraMigrate::ContentMover do
     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 }
+      it { is_expected.to be nil }
     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
+      it { is_expected.to eql "There was a problem with sparql 404 Error!" }
     end
   end
 

data/spec/unit/datastream_verification_spec.rb

@@ -23,10 +23,7 @@ describe FedoraMigrate::DatastreamVerification do
     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
+      it      { is_expected.to_not be_valid }
     end
     context "when the checksum is missing" do
       subject { TestSubject.new(missing_checksum) }
@@ -37,10 +34,7 @@ describe FedoraMigrate::DatastreamVerification do
       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
+        it     { is_expected.to_not be_valid }
       end
     end
   end

data/spec/unit/migration_report_spec.rb

@@ -0,0 +1,58 @@
+require 'spec_helper'
+
+describe FedoraMigrate::MigrationReport do
+
+  let(:existing_report) { FedoraMigrate::MigrationReport.new("spec/fixtures/sample-report.json") }
+  let(:new_report)      { FedoraMigrate::MigrationReport.new }
+
+  context "with an existing report" do
+    subject { existing_report }
+    it { is_expected.not_to be_empty }
+    describe "::results" do
+      subject { existing_report.results }
+      it { is_expected.to be_kind_of(Hash) }
+    end
+    describe "::failed_objects" do
+      subject { existing_report.failed_objects }
+      it { is_expected.to include("scholarsphere:6395wb555", "scholarsphere:x346dm27k") }
+    end
+    describe "::failures" do
+      subject { existing_report.failures }    
+      context "when the report contains failed migrations" do
+        it { is_expected.to eq 2 }
+      end
+    end
+    describe "::total_objects" do
+      subject { existing_report.total_objects }
+      it { is_expected.to eq 5 }
+    end
+    describe "::report_failures" do
+      subject { existing_report.report_failures }   
+      it { is_expected.to be_kind_of(String) }
+    end
+    describe "::save" do
+      context "with the default path" do
+        it "should write the report" do
+          expect(File).to receive(:write).with("report.json", "{\n}")
+          new_report.save
+        end
+      end
+      context "with a user-provided path" do
+        it "should write the report" do
+          expect(File).to receive(:write).with("foo/path/report.json", "{\n}")
+          new_report.save("foo/path")
+        end
+      end
+    end
+  end
+
+  context "as a new report" do
+    subject { new_report }
+    it { is_expected.to be_empty }
+    describe "::results" do
+      subject { new_report.results }
+      it { is_expected.to be_kind_of(Hash) }
+    end
+  end
+
+end

data/spec/unit/object_mover_spec.rb

@@ -14,7 +14,7 @@ describe FedoraMigrate::ObjectMover do
 
   describe "#prepare_target" do
     subject do
-      FedoraMigrate::ObjectMover.new("source", "target").prepare_target
+      FedoraMigrate::ObjectMover.new("source", double("Target", id: nil)).prepare_target
     end
     it "should call the before hook and save the target" do
       expect_any_instance_of(FedoraMigrate::ObjectMover).to receive(:before_object_migration)
@@ -24,7 +24,7 @@ describe FedoraMigrate::ObjectMover do
 
   describe "#complete_target" do
     subject do
-      FedoraMigrate::ObjectMover.new("source", "target").complete_target
+      FedoraMigrate::ObjectMover.new("source", double("Target", id: nil)).complete_target
     end
     it "should call the after hook and save the target" do
       expect_any_instance_of(FedoraMigrate::ObjectMover).to receive(:after_object_migration)

data/spec/unit/repository_migrator_spec.rb

@@ -5,23 +5,52 @@ describe FedoraMigrate::RepositoryMigrator do
   let(:namespace) { "sufia" }
 
   it { is_expected.to respond_to(:source_objects) }
-  it { is_expected.to respond_to(:failed) }
+  it { is_expected.to respond_to(:report) }
   it { is_expected.to respond_to(:namespace) }
 
-  describe "#failed" do
-    specify "sets to zero" do
-      expect(subject.failed).to eql(0)
+  def mock_report content
+    report = FedoraMigrate::MigrationReport.new 
+    report.results = JSON.parse(content.to_json)
+    report
+  end
+
+  describe "#failures" do
+    context "when objects have failed to migrate" do
+      let(:failing_report) { { "sufia:rb68xc089" => FedoraMigrate::RepositoryMigrator::SingleObjectReport.new(false, "objects", "relationships") } }
+      before { allow_any_instance_of(FedoraMigrate::RepositoryMigrator).to receive(:report).and_return(mock_report(failing_report)) }
+      subject do
+        migrator = FedoraMigrate::RepositoryMigrator.new(namespace)
+        migrator.failures
+      end
+      it { is_expected.to be 1 }
+    end
+    context "when all objects have migrated" do
+      let(:passing_report) { { "sufia:rb68xc089" => FedoraMigrate::RepositoryMigrator::SingleObjectReport.new(true, "objects", "relationships") } }
+      before { allow_any_instance_of(FedoraMigrate::RepositoryMigrator).to receive(:report).and_return(mock_report(passing_report)) }
+      subject do
+        migrator = FedoraMigrate::RepositoryMigrator.new(namespace)
+        migrator.failures
+      end
+      it { is_expected.to eql 0 }
     end
   end
 
-  context "when forcing" do
+  describe "forcing relationship migration" 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) 
+      allow(subject).to receive(:source_objects).and_return([])
+      allow(subject).to receive(:failures).and_return(1)
+    end
+    context "without an explicit force" do
+      subject { FedoraMigrate::RepositoryMigrator.new(namespace) }
+      it "should not migrate relationships" do
+        expect(subject.migrate_relationships).to eql("Relationship migration halted because 1 objects didn't migrate successfully.")
+      end
     end
-    subject { FedoraMigrate::RepositoryMigrator.new(namespace, { force: true }) }
-    specify "migrate relationships if failures are present" do
-      expect(subject.migrate_relationships).to be true
+    context "with an explicit force" do
+      subject { FedoraMigrate::RepositoryMigrator.new(namespace, { force: true }) }
+      it "should migrate relationships" do
+        expect(subject.migrate_relationships).to_not be_nil
+      end
     end
   end