iiif_print 1.1.0 → 2.0.0

data/spec/iiif_print/metadata_spec.rb

@@ -1,249 +0,0 @@
-require 'spec_helper'
-
-RSpec.describe IiifPrint::Metadata do
-  let(:base_url) { "https://my.dev.test" }
-  let(:solr_hit) { SolrHit.new(attributes) }
-  let(:fields) { IiifPrint.default_fields(fields: metadata_fields) }
-  let(:metadata_fields) do
-    {
-      title: {},
-      description: {},
-      date_modified: {}
-    }
-  end
-
-  describe ".build_metadata_for" do
-    subject(:manifest_metadata) do
-      described_class.build_metadata_for(
-        work: solr_hit,
-        version: version,
-        fields: fields,
-        current_ability: double(Ability),
-        base_url: base_url
-      )
-    end
-
-    context "for version 2 of the IIIF spec" do
-      let(:version) { 2 }
-
-      context "with a field that has some plain text" do
-        let(:attributes) { { "title_tesim" => ["My Awesome Title"] } }
-
-        it "maps the metadata accordingly" do
-          expect(manifest_metadata).to eq [
-            { "label" => "Title", "value" => ["My Awesome Title"] }
-          ]
-        end
-      end
-
-      context "with a field that contains a url string" do
-        let(:attributes) { { "description_tesim" => ["A url like https://www.example.com/, cool!"] } }
-
-        it "creates a link for the url string" do
-          expect(manifest_metadata).to eq [
-            { "label" => "Description",
-               "value" =>
-                [
-                  "A url like <a href='https://www.example.com/' target='_blank'>https://www.example.com/</a>, cool!"
-                ] }
-          ]
-        end
-      end
-
-      context "with a date" do
-        let(:attributes) { { "date_modified_dtsi" => "2011-11-11T11:11:11Z" } }
-
-        it "displays it just the date" do
-          expect(manifest_metadata).to eq [{ "label" => "Date modified", "value" => ["2011-11-11"] }]
-        end
-      end
-
-      context "with a faceted option" do
-        let(:metadata_fields) { { creator: { render_as: :faceted } } }
-        let(:attributes) { { "creator_tesim" => ["McAuthor, Arthur"] } }
-
-        it "adds a link to the faceted search" do
-          expect(manifest_metadata).to eq [
-            { "label" => "Creator",
-              "value" =>
-                ["<a href='#{base_url}/catalog?f%5Bcreator_sim%5D%5B%5D=McAuthor%2C+Arthur&locale=en'>McAuthor, Arthur</a>"] }
-          ]
-        end
-      end
-
-      context "with an authority option" do
-        context "rights statement" do
-          let(:metadata_fields) { { rights_statement: { render_as: :rights_statement } } }
-          let(:attributes) { { "rights_statement_tesim" => ["http://rightsstatements.org/vocab/InC-OW-EU/1.0/"] } }
-
-          it "renders a link and displays a term" do
-            expect(manifest_metadata).to eq [
-              { "label" => "Rights statement",
-                "value" => ["<a href='http://rightsstatements.org/vocab/InC-OW-EU/1.0/'>In Copyright - EU Orphan Work</a>"] }
-            ]
-          end
-        end
-
-        context "license" do
-          let(:metadata_fields) { { license: { render_as: :license } } }
-          let(:attributes) { { "license_tesim" => ["https://creativecommons.org/licenses/by-sa/4.0/"] } }
-
-          it "renders a link and displays a term" do
-            expect(manifest_metadata).to eq [
-              { "label" => "License",
-                "value" => [
-                  "<a href='https://creativecommons.org/licenses/by-sa/4.0/'>Creative Commons BY-SA Attribution-ShareAlike 4.0 International</a>"
-                ] }
-            ]
-          end
-        end
-      end
-
-      context "when the work is apart of a collection" do
-        let(:metadata_fields) { { collection: {} } }
-        let(:collection_attributes) { { "id" => "321cba", "title_tesim" => ["My Cool Collection"] } }
-        let(:collection_solr_doc) { SolrDocument.new(collection_attributes) }
-        let(:attributes) { { "member_of_collection_ids_ssim" => "321cba" } }
-
-        it "renders a link to the collection" do
-          allow(SolrDocument).to receive(:find)
-          allow(Hyrax::CollectionMemberService).to receive(:run).and_return([collection_solr_doc])
-          expect(manifest_metadata).to eq [
-            { "label" => "Collection",
-              "value" => ["<a href='#{base_url}/collections/321cba'>My Cool Collection</a>"] }
-          ]
-        end
-      end
-
-      context "when the value has an empty string" do
-        let(:attributes) { { "title_tesim" => ["This is a title."], "description_tesim" => [""] } }
-
-        it "does not map the field with an empty string" do
-          expect(manifest_metadata.flat_map(&:values)).not_to include([""])
-          expect(manifest_metadata).to eq [{ "label" => "Title", "value" => ["This is a title."] }]
-        end
-      end
-
-      context "when the value is an empty string" do
-        let(:attributes) { { "description_tesim" => [""] } }
-
-        it "returns and empty array" do
-          expect(manifest_metadata).to eq []
-        end
-      end
-    end
-
-    context "for version 3 of the IIIF spec" do
-      let(:version) { 3 }
-
-      context "with a field that has some plain text" do
-        let(:attributes) { { "title_tesim" => ["My Awesome Title"] } }
-
-        # NOTE: this assumes the I18n.locale is set as :en
-        it "maps the metadata accordingly" do
-          expect(manifest_metadata).to eq [{ "label" => { "en" => ["Title"] },
-                                             "value" => { "none" => ["My Awesome Title"] } }]
-        end
-      end
-
-      context "with a field that contains a url string" do
-        let(:attributes) { { "description_tesim" => ["A url like https://www.example.com/, cool!"] } }
-
-        it "creates a link for the url string" do
-          expect(manifest_metadata).to eq [
-            { "label" => { "en" => ["Description"] },
-              "value" => { "none" =>
-                ["A url like <a href='https://www.example.com/' target='_blank'>https://www.example.com/</a>, cool!"] } }
-          ]
-        end
-      end
-
-      context "with a date" do
-        let(:attributes) { { "date_modified_dtsi" => "2011-11-11T11:11:11Z" } }
-
-        it "displays it just the date" do
-          expect(manifest_metadata).to eq [{ "label" => { "en" => ["Date modified"] },
-                                             "value" => { "none" => ["2011-11-11"] } }]
-        end
-      end
-
-      context "with a faceted option" do
-        let(:metadata_fields) { { creator: { render_as: :faceted } } }
-        let(:attributes) { { "creator_tesim" => ["McAuthor, Arthur"] } }
-
-        it "adds a link to the faceted search" do
-          expect(manifest_metadata). to eq [
-            { "label" => { "en" => ["Creator"] },
-              "value" => { "none" =>
-                ["<a href='#{base_url}/catalog?f%5Bcreator_sim%5D%5B%5D=McAuthor%2C+Arthur&locale=en'>McAuthor, Arthur</a>"] } }
-          ]
-        end
-      end
-
-      context "with an authority option" do
-        context "rights statement" do
-          let(:metadata_fields) { { rights_statement: { render_as: :rights_statement } } }
-          let(:attributes) { { "rights_statement_tesim" => ["http://rightsstatements.org/vocab/InC-OW-EU/1.0/"] } }
-
-          it "renders a link and displays a term" do
-            expect(manifest_metadata).to eq [
-              { "label" => { "en" => ["Rights statement"] },
-                "value" => { "none" => [
-                  "<a href='http://rightsstatements.org/vocab/InC-OW-EU/1.0/'>In Copyright - EU Orphan Work</a>"
-                ] } }
-            ]
-          end
-        end
-
-        context "license" do
-          let(:metadata_fields) { { license: { render_as: :license } } }
-          let(:attributes) { { "license_tesim" => ["https://creativecommons.org/licenses/by-sa/4.0/"] } }
-
-          it "renders a link and displays a term" do
-            expect(manifest_metadata).to eq [
-              { "label" => { "en" => ["License"] },
-                "value" => { "none" => [
-                  "<a href='https://creativecommons.org/licenses/by-sa/4.0/'>Creative Commons BY-SA Attribution-ShareAlike 4.0 International</a>"
-                ] } }
-            ]
-          end
-        end
-      end
-
-      context "when the work is apart of a collection" do
-        let(:metadata_fields) { { collection: {} } }
-        let(:collection_attributes) { { "id" => "321cba", "title_tesim" => ["My Cool Collection"] } }
-        let(:collection_solr_doc) { SolrDocument.new(collection_attributes) }
-        let(:attributes) { { "member_of_collection_ids_ssim" => "321cba" } }
-
-        it "renders a link to the collection" do
-          allow(SolrDocument).to receive(:find)
-          allow(Hyrax::CollectionMemberService).to receive(:run).and_return([collection_solr_doc])
-          expect(manifest_metadata).to eq [
-            { "label" => { "en" => ["Collection"] },
-              "value" => { "none" => ["<a href='#{base_url}/collections/321cba'>My Cool Collection</a>"] } }
-          ]
-        end
-      end
-
-      context "when the value has an empty string" do
-        let(:attributes) { { "title_tesim" => ["This is a title."], "description_tesim" => [""] } }
-
-        it "does not map the field with an empty string" do
-          expect(manifest_metadata.flat_map(&:values)).not_to include({ "none" => [""] })
-          expect(manifest_metadata).to eq [
-            { "label" => { "en" => ["Title"] }, "value" => { "none" => ["This is a title."] } }
-          ]
-        end
-      end
-
-      context "when the value is an empty string" do
-        let(:attributes) { { "description_tesim" => [""] } }
-
-        it "returns and empty array" do
-          expect(manifest_metadata).to eq []
-        end
-      end
-    end
-  end
-end

data/spec/iiif_print/split_pdfs/base_splitter_spec.rb

@@ -1,27 +0,0 @@
-require 'spec_helper'
-
-RSpec.describe IiifPrint::SplitPdfs::BaseSplitter do
-  let(:path) { __FILE__ }
-  let(:splitter) { described_class.new(path) }
-  subject { described_class }
-
-  it { is_expected.to respond_to(:call) }
-
-  describe "instance" do
-    subject { splitter }
-
-    it { is_expected.to respond_to :compression }
-    it { is_expected.to respond_to :compression? }
-    it { is_expected.to respond_to :image_extension }
-    it { is_expected.to respond_to :quality }
-  end
-
-  describe '#compression' do
-    it 'can be changed within the instance' do
-      expect do
-        splitter.compression = 'squishy'
-      end.not_to change(splitter.class, :compression)
-      expect(splitter.compression).to eq('squishy')
-    end
-  end
-end

data/spec/iiif_print/split_pdfs/derivative_rodeo_splitter_spec.rb

@@ -1,80 +0,0 @@
-# frozen_string_literal: true
-
-require 'spec_helper'
-
-RSpec.describe IiifPrint::SplitPdfs::DerivativeRodeoSplitter do
-  let(:filename) { __FILE__ }
-  let(:work) { double(MyWork, id: 'id-12345', aark_id: '12345') }
-  let(:file_set) { FileSet.new.tap { |fs| fs.save!(validate: false) } }
-  let(:location_stub) { double(DerivativeRodeo::StorageLocations::BaseLocation, exist?: true) }
-
-  before do
-    allow(DerivativeRodeo::StorageLocations::BaseLocation).to receive(:from_uri).and_return(location_stub)
-  end
-
-  describe 'class' do
-    subject { described_class }
-
-    it { is_expected.to respond_to(:call) }
-  end
-
-  subject(:instance) { described_class.new(filename, file_set: file_set, output_tmp_dir: Dir.tmpdir) }
-  let(:generator) { double(DerivativeRodeo::Generators::PdfSplitGenerator, generated_files: []) }
-
-  before do
-    allow(file_set).to receive(:parent).and_return(work)
-    # TODO: This is a hack that leverages the internals of Hydra::Works; not excited about it but
-    # this part is only one piece of the over all integration.
-    allow(file_set).to receive(:original_file).and_return(double(original_filename: __FILE__))
-  end
-
-  it { is_expected.to respond_to :split_files }
-
-  it 'uses the rodeo to split' do
-    expect(DerivativeRodeo::Generators::PdfSplitGenerator).to receive(:new).and_return(generator)
-    described_class.call(filename, file_set: file_set)
-  end
-
-  describe '#preprocessed_location_template' do
-    let(:derivative_rodeo_preprocessed_file) { IiifPrint::DerivativeRodeoService.derivative_rodeo_uri(file_set: file_set, filename: filename) }
-    let(:import_url) { "https://somewhere.com/that/exists.pdf" }
-    subject { instance.preprocessed_location_template }
-
-    context 'when the s3 file exists in the rodeo' do
-      it 'is that file' do
-        is_expected.to eq(derivative_rodeo_preprocessed_file)
-      end
-    end
-
-    context 'when the s3 file does not exist in the rodeo and we have the local file' do
-      it 'is the import_url' do
-        expect_any_instance_of(DerivativeRodeo::Generators::CopyGenerator).not_to receive(:generated_uris)
-        file_set.import_url = import_url
-        expect(instance).to receive(:rodeo_conformant_uri_exists?).with(derivative_rodeo_preprocessed_file).and_return(false)
-        expect(instance).to receive(:rodeo_conformant_uri_exists?).with(instance.input_uri).and_return(true)
-        expect(subject).to eq(instance.input_uri)
-      end
-    end
-
-    context 'when the s3 file does not exist and we do not have the input URI nor the given import url does NOT exist' do
-      let(:generator) { double(DerivativeRodeo::Generators::CopyGenerator, generated_uris: ["file:///generated/uri"]) }
-      it 'will invoke the DerivativeRodeo::Generators::CopyGenerator to bring the file locally' do
-        allow(DerivativeRodeo::Generators::CopyGenerator).to receive(:new).and_return(generator)
-        file_set.import_url = import_url
-        expect(instance).to receive(:rodeo_conformant_uri_exists?).with(derivative_rodeo_preprocessed_file).and_return(false)
-        expect(instance).to receive(:rodeo_conformant_uri_exists?).with(instance.input_uri).and_return(false)
-
-        expect(subject).to eq(generator.generated_uris.first)
-      end
-    end
-
-    context "when the s3 file does not exist and we don't have a remote_url" do
-      it 'will use the given filename' do
-        file_set.import_url = nil
-        expect(instance).to receive(:rodeo_conformant_uri_exists?).with(derivative_rodeo_preprocessed_file).and_return(false)
-
-        expect(subject).to eq(nil)
-      end
-    end
-  end
-end

data/spec/iiif_print/split_pdfs/destroy_pdf_child_works_service_spec.rb

@@ -1,92 +0,0 @@
-# frozen_string_literal: true
-
-require 'spec_helper'
-
-RSpec.describe IiifPrint::SplitPdfs::DestroyPdfChildWorksService do
-  let(:subject) { described_class.conditionally_destroy_spawned_children_of(file_set: fileset, work: work) }
-
-  let(:work) { WorkWithIiifPrintConfig.new(title: ['required title'], id: '123') }
-  let(:fileset) { FileSet.new.tap { |fs| fs.save!(validate: false) } }
-  let(:child_work) { WorkWithIiifPrintConfig.new(title: ["Child of #{work.id} file.pdf page 01"], id: '456', is_child: true) }
-  let(:pending_rel1) do
-    IiifPrint::PendingRelationship.new(
-    parent_id: work.id,
-    child_title: "Child of #{work.id} file.pdf page 01",
-    child_order: "Child of #{work.id} file.pdf page 01",
-    parent_model: WorkWithIiifPrintConfig,
-    child_model: WorkWithIiifPrintConfig,
-    file_id: fileset.id
-  )
-  end
-  let(:pending_rel2) do
-    IiifPrint::PendingRelationship.new(
-    parent_id: work.id,
-    child_title: "Child of #{work.id} another.pdf page 01",
-    child_order: "Child of #{work.id} another.pdf page 01",
-    parent_model: WorkWithIiifPrintConfig,
-    child_model: WorkWithIiifPrintConfig,
-    file_id: 'another'
-  )
-  end
-  # let(:uploaded_pdf_file) { create(:uploaded_pdf_file) }
-  # let(:uploaded_file_ids) { [uploaded_pdf_file.id] }
-
-  before do
-    allow(fileset).to receive(:parent).and_return(work)
-    allow(fileset).to receive(:label).and_return('file.pdf')
-    allow(fileset).to receive(:mime_type).and_return('application/pdf')
-  end
-
-  describe 'class' do
-    subject { described_class }
-
-    it { is_expected.to respond_to(:conditionally_destroy_spawned_children_of) }
-    it { is_expected.not_to respond_to(:destroy_spawned_children) }
-  end
-
-  describe '#conditionally_destroy_spawned_children_of' do
-    context 'with child works by fileset id' do
-      before do
-        allow(WorkWithIiifPrintConfig).to receive(:where).with(split_from_pdf_id: fileset.id).and_return([child_work])
-      end
-
-      it 'destroys the child works' do
-        expect(child_work).to receive(:destroy)
-        subject
-      end
-    end
-
-    context 'with child works by title' do
-      before do
-        allow(WorkWithIiifPrintConfig).to receive(:where).with(split_from_pdf_id: fileset.id).and_return([])
-        allow(WorkWithIiifPrintConfig).to receive(:where).and_return([child_work])
-      end
-
-      it 'destroys the child works' do
-        expect(child_work).to receive(:destroy)
-        subject
-      end
-    end
-
-    context 'when fileset is not a PDF mimetype' do
-      before do
-        allow(fileset).to receive(:mime_type).and_return('not_pdf')
-      end
-
-      it 'returns with no changes' do
-        expect(IiifPrint::PendingRelationship).not_to receive(:where)
-      end
-    end
-
-    context 'when IiifPrint::PendingRelationship records exist' do
-      before do
-        pending_rel1.save
-        pending_rel2.save
-      end
-
-      it 'deletes only records associated with the specific fileset PDF file' do
-        expect { subject }.to change(IiifPrint::PendingRelationship, :count).by(-1)
-      end
-    end
-  end
-end

data/spec/iiif_print/split_pdfs/pages_to_jpgs_splitter_spec.rb

@@ -1,22 +0,0 @@
-require 'spec_helper'
-require 'misc_shared'
-
-RSpec.describe IiifPrint::SplitPdfs::PagesToJpgsSplitter do
-  let(:path) { __FILE__ }
-  let(:splitter) { described_class.new(path) }
-
-  describe '#quality' do
-    subject { splitter.quality }
-    it { is_expected.to eq(described_class.quality) }
-  end
-
-  describe '#quality?' do
-    subject { splitter.quality? }
-    it { is_expected.to be_truthy }
-  end
-
-  describe '#image_extension' do
-    subject { splitter.image_extension }
-    it { is_expected.to eq('jpg') }
-  end
-end

data/spec/iiif_print/split_pdfs/pages_to_pngs_splitter_spec.rb

@@ -1,18 +0,0 @@
-require 'spec_helper'
-
-RSpec.describe IiifPrint::SplitPdfs::PagesToPngsSplitter do
-  describe '.compression' do
-    subject { described_class.compression }
-    it { is_expected.to be_nil }
-  end
-
-  describe '.compression?' do
-    subject { described_class.compression? }
-    it { is_expected.to be_falsey }
-  end
-
-  describe '.image_extension' do
-    subject { described_class.image_extension }
-    it { is_expected.to eq('png') }
-  end
-end

data/spec/iiif_print/split_pdfs/pages_to_tiffs_splitter_spec.rb

@@ -1,19 +0,0 @@
-require 'spec_helper'
-require 'misc_shared'
-
-RSpec.describe IiifPrint::SplitPdfs::PagesToTiffsSplitter do
-  describe '.compression' do
-    subject { described_class.compression }
-    it { is_expected.to eq(described_class::DEFAULT_COMPRESSION) }
-  end
-
-  describe '.compression?' do
-    subject { described_class.compression? }
-    it { is_expected.to be_truthy }
-  end
-
-  describe '.image_extension' do
-    subject { described_class.image_extension }
-    it { is_expected.to eq('tiff') }
-  end
-end

data/spec/iiif_print/text_extraction/alto_reader_spec.rb

@@ -1,49 +0,0 @@
-require 'json'
-require 'spec_helper'
-
-RSpec.describe IiifPrint::TextExtraction::AltoReader do
-  let(:fixture_path) do
-    File.join(
-      IiifPrint::GEM_PATH, 'spec', 'fixtures', 'files'
-    )
-  end
-
-  let(:minimal_path) { File.join(fixture_path, 'minimal-alto.xml') }
-  let(:ndnp_alto_path) { File.join(fixture_path, 'ndnp-alto-sample.xml') }
-  let(:minimal) { File.read(minimal_path) }
-
-  let(:reader_minimal) { described_class.new(minimal) }
-  let(:reader_minimal_path) { described_class.new(minimal_path) }
-  let(:reader_ndnp) { described_class.new(ndnp_alto_path) }
-
-  describe "reads alto" do
-    it "loads ALTO source" do
-      expect(reader_minimal_path.source).to eq reader_minimal.source
-      expect(reader_minimal_path.source.size).to eq 1383
-      expect(reader_ndnp.source.size).to eq 1_050_876
-    end
-
-    it "loads document stream" do
-      expect(reader_minimal_path.doc_stream).to be_kind_of Nokogiri::XML::SAX::Document
-      expect(reader_minimal_path.doc_stream).to respond_to :text
-      expect(reader_minimal_path.doc_stream).to respond_to :words
-    end
-  end
-
-  describe "outputs text derivative formats" do
-    it "outputs plain text" do
-      # try simple flat text input
-      expect(reader_minimal.text).to eq "This is only a test."
-      expect(reader_minimal.text).to eq reader_minimal.doc_stream.text
-      # try more complex input
-      expect(reader_ndnp.text.size).to eq 30_519
-    end
-
-    it "passes args to WordCoordsBuilder and receives output" do
-      parsed = JSON.parse(reader_minimal.json)
-      expect(parsed['coords'].length).to be > 1
-      parsed = JSON.parse(reader_ndnp.json)
-      expect(parsed['coords'].size).to eq 2_125
-    end
-  end
-end

data/spec/iiif_print/text_extraction/hocr_reader_spec.rb

@@ -1,45 +0,0 @@
-require 'json'
-require 'nokogiri'
-require 'spec_helper'
-
-RSpec.describe IiifPrint::TextExtraction::HOCRReader do
-  let(:fixture_path) do
-    File.join(
-      IiifPrint::GEM_PATH, 'spec', 'fixtures', 'files'
-    )
-  end
-
-  let(:minimal_path) { File.join(fixture_path, 'ocr_mono_text_hocr.html') }
-  let(:minimal) { File.read(minimal_path) }
-
-  let(:reader_minimal) { described_class.new(minimal) }
-  let(:reader_minimal_path) { described_class.new(minimal_path) }
-
-  describe "reads hOCR" do
-    it "loads hOCR either from path or source text" do
-      expect(reader_minimal_path.source).to eq reader_minimal.source
-      # size here is in Unicode characters, not bytes:
-      expect(reader_minimal_path.source.size).to eq 16_590
-    end
-
-    it "loads document stream" do
-      expect(reader_minimal_path.doc_stream).to be_kind_of Nokogiri::XML::SAX::Document
-      expect(reader_minimal_path.doc_stream).to respond_to :text
-      expect(reader_minimal_path.doc_stream).to respond_to :words
-    end
-  end
-
-  describe "outputs text derivative formats" do
-    it "outputs plain text" do
-      plain_text = reader_minimal.text
-      expect(plain_text.slice(0, 40)).to eq "_A FEARFUL ADVENTURE.\n‘The Missouri. Rep"
-      expect(reader_minimal.text).to eq reader_minimal.doc_stream.text
-      expect(reader_minimal.text.size).to eq 723
-    end
-
-    it "passes args to WordCoordsBuilder and receives output" do
-      parsed = JSON.parse(reader_minimal.json)
-      expect(parsed['coords'].length).to be > 1
-    end
-  end
-end