iiif_print 1.0.0

data/spec/iiif_print/text_extraction/page_ocr_spec.rb

@@ -0,0 +1,84 @@
+require 'json'
+require 'nokogiri'
+require 'spec_helper'
+
+RSpec.describe IiifPrint::TextExtraction::PageOCR do
+  let(:fixture_path) do
+    File.join(
+      IiifPrint::GEM_PATH, 'spec', 'fixtures', 'files'
+    )
+  end
+
+  let(:altoxsd) do
+    xsdpath = File.join(fixture_path, 'alto-2-0.xsd')
+    Nokogiri::XML::Schema(File.read(xsdpath))
+  end
+
+  # sample "snippet" images for OCR testing:
+  let(:example_gray_tiff) { File.join(fixture_path, 'ocr_gray.tiff') }
+  let(:example_mono_tiff) { File.join(fixture_path, 'ocr_mono.tiff') }
+  let(:example_color_tiff) { File.join(fixture_path, 'ocr_color.tiff') }
+  let(:example_gray_jp2) { File.join(fixture_path, 'ocr_gray.jp2') }
+  let(:ocr_from_gray_tiff) { described_class.new(example_gray_tiff) }
+
+  describe "performs OCR" do
+    def match_ocr_expectations(words)
+      expect(words).to be_an(Array)
+      expect(words).not_to be_empty
+      expect(words[0]).to be_a(Hash)
+      [:word, :coordinates].each do |key|
+        expect(words[0].keys).to include key
+      end
+    end
+
+    it "gets words and coordinates from grayscale source" do
+      match_ocr_expectations(ocr_from_gray_tiff.words)
+    end
+
+    it "gets words and coordinates from one-bit source" do
+      ocr = described_class.new(example_mono_tiff)
+      match_ocr_expectations(ocr.words)
+    end
+
+    it "gets words and coordinates from color source" do
+      ocr = described_class.new(example_color_tiff)
+      match_ocr_expectations(ocr.words)
+    end
+
+    it "gets words and coordinates from jp2 source" do
+      ocr = described_class.new(example_gray_jp2)
+      match_ocr_expectations(ocr.words)
+    end
+  end
+
+  describe "turns image into ALTO" do
+    xit "takes grayscale tiff, outputs valid ALTO, geometry" do
+      alto = ocr_from_gray_tiff.alto
+      document = Nokogiri::XML(alto)
+      errors = altoxsd.validate(document)
+      expect(errors.length).to eq 0
+      expect(document.at_css('PrintSpace')['WIDTH']).to eq "418"
+      expect(document.at_css('PrintSpace')['HEIGHT']).to eq "1046"
+    end
+  end
+
+  describe "plain text" do
+    it "makes plain text available for image" do
+      plain = ocr_from_gray_tiff.plain
+      expect(plain.class).to be String
+      expect(plain.length).to be > 0
+    end
+  end
+
+  describe "JSON word coordinates" do
+    it "passes properly formatted data to WordCoordsBuilder and receives output" do
+      parsed = JSON.parse(ocr_from_gray_tiff.word_json)
+      expect(parsed['coords'].length).to be > 1
+      word = ocr_from_gray_tiff.words[0]
+      word1 = parsed['coords'][word[:word]]
+      word1_coords = word1[0]
+      expect(word1_coords[2]).to eq word[:coordinates][2]
+      expect(word1_coords[3]).to eq word[:coordinates][3]
+    end
+  end
+end

data/spec/iiif_print/text_extraction/render_alto_spec.rb

@@ -0,0 +1,54 @@
+require 'spec_helper'
+
+RSpec.describe IiifPrint::TextExtraction::RenderAlto do
+  let(:fixture_path) do
+    File.join(
+      IiifPrint::GEM_PATH, 'spec', 'fixtures', 'files'
+    )
+  end
+
+  let(:altoxsd) do
+    xsdpath = File.join(fixture_path, 'alto-2-0.xsd')
+    Nokogiri::XML::Schema(File.read(xsdpath))
+  end
+
+  let(:page_prefix) { '<Page ID="ID1" PHYSICAL_IMG_NR="1"' }
+
+  let(:words) do
+    [
+      { word: "If", coordinates: [52, 13, 11, 14] },
+      { word: "you", coordinates: [69, 17, 31, 14] },
+      { word: "are", coordinates: [108, 17, 28, 10] },
+      { word: "a", coordinates: [143, 17, 8, 10] },
+      { word: "friend,", coordinates: [158, 13, 56, 16] },
+      { word: "you", coordinates: [51, 39, 31, 14] },
+      { word: "speak", coordinates: [90, 35, 50, 18] },
+      { word: "the", coordinates: [146, 35, 28, 14] },
+      { word: "password,", coordinates: [182, 35, 85, 18] },
+      { word: "and", coordinates: [51, 57, 30, 14] },
+      { word: "the", coordinates: [89, 57, 28, 14] },
+      { word: "doors", coordinates: [124, 57, 48, 14] },
+      { word: "will", coordinates: [180, 57, 28, 14] },
+      { word: "open.", coordinates: [216, 61, 47, 14] }
+    ]
+  end
+
+  describe "renders alto" do
+    it "creates alto given width, height, words" do
+      renderer = described_class.new(12_000, 9600)
+      output = renderer.to_alto(words)
+      expect(output.class).to be String
+      expect(output).to include '<alto'
+      expect(output).to include '<String'
+      expect(output).to include page_prefix + ' HEIGHT="9600" WIDTH="12000"'
+      expect(Nokogiri::XML(output).errors.empty?).to be true
+    end
+
+    xit "makes alto 2.0 that validates" do
+      renderer = described_class.new(12_000, 9600)
+      output = renderer.to_alto(words)
+      document = Nokogiri::XML(output)
+      altoxsd.validate(document)
+    end
+  end
+end

data/spec/iiif_print/text_extraction/word_coords_builder_spec.rb

@@ -0,0 +1,44 @@
+require 'spec_helper'
+
+RSpec.describe IiifPrint::TextExtraction::WordCoordsBuilder do
+  let(:words) do
+    [
+      { word: "foo", coordinates: [1, 2, 3, 4] },
+      { word: "bar", coordinates: [5, 6, 7, 8] },
+      { word: "baz", coordinates: [9, 10, 11, 12] },
+      { word: "foo", coordinates: [13, 14, 15, 16] }
+    ]
+  end
+  let(:image_width) { 1_234 }
+  let(:image_height) { 5_678 }
+
+  describe '.json_coordinates_for' do
+    let(:wcb_to_json) { JSON.parse(described_class.json_coordinates_for(words: words, width: image_width, height: image_height)) }
+    it 'has the correct structure' do
+      expect(wcb_to_json['height']).to eq image_height
+      expect(wcb_to_json['width']).to eq image_width
+      expect(wcb_to_json['coords'].length).to eq 3
+      expect(wcb_to_json['coords']['foo']).not_to be_falsey
+    end
+
+    it 'combines coordinates for the same word' do
+      expect(wcb_to_json['coords']['foo']).to eq [[1, 2, 3, 4], [13, 14, 15, 16]]
+    end
+  end
+
+  describe '#to_json' do
+    let(:wcb_to_json) { JSON.parse(wcb.to_json) }
+    let(:wcb) { described_class.new(words, image_width, image_height) }
+
+    it 'has the correct structure' do
+      expect(wcb_to_json['height']).to eq image_height
+      expect(wcb_to_json['width']).to eq image_width
+      expect(wcb_to_json['coords'].length).to eq 3
+      expect(wcb_to_json['coords']['foo']).not_to be_falsey
+    end
+
+    it 'combines coordinates for the same word' do
+      expect(wcb_to_json['coords']['foo']).to eq [[1, 2, 3, 4], [13, 14, 15, 16]]
+    end
+  end
+end

data/spec/iiif_print_spec.rb

@@ -0,0 +1,51 @@
+require 'spec_helper'
+
+RSpec.describe IiifPrint do
+  describe ".manifest_metadata_for" do
+    let(:attributes) do
+      { "id" => "abc123",
+        "title_tesim" => ['My Awesome Title'] }
+    end
+    let(:solr_document) { SolrDocument.new(attributes) }
+    let(:base_url) { "https://my.dev.test" }
+
+    subject(:manifest_metadata) do
+      described_class.manifest_metadata_for(work: solr_document, current_ability: double(Ability), base_url: base_url)
+    end
+    it { is_expected.not_to be_falsey }
+    it "does not contain any nil values" do
+      expect(subject).not_to include(nil)
+    end
+  end
+
+  describe ".model_configuration" do
+    context "default configuration" do
+      let(:model) do
+        Class.new do
+          include IiifPrint.model_configuration(pdf_split_child_model: Class.new)
+        end
+      end
+
+      subject(:record) { model.new }
+
+      it { is_expected.to be_iiif_print_config }
+
+      it "has a #pdf_splitter_job" do
+        expect(record.iiif_print_config.pdf_splitter_job).to be(IiifPrint::Jobs::ChildWorksFromPdfJob)
+      end
+
+      it "has a #pdf_splitter_service" do
+        expect(record.iiif_print_config.pdf_splitter_service).to be(IiifPrint::SplitPdfs::PagesIntoImagesService)
+      end
+
+      it "has #derivative_service_plugins" do
+        expect(record.iiif_print_config.derivative_service_plugins).to eq(
+          [IiifPrint::JP2DerivativeService,
+           IiifPrint::PDFDerivativeService,
+           IiifPrint::TextExtractionDerivativeService,
+           IiifPrint::TIFFDerivativeService]
+        )
+      end
+    end
+  end
+end

data/spec/misc_shared.rb

@@ -0,0 +1,111 @@
+RSpec.shared_context "shared setup", shared_context: :metadata do
+  let(:fixture_path) do
+    path = File.join(
+      IiifPrint::GEM_PATH, 'spec', 'fixtures', 'files'
+    )
+    # TODO: NOTE: this has potential timing issues in the specs, because we're adjusting the
+    # configured value during the spec run.
+    registered = Hyrax.config.registered_ingest_dirs
+    registered.push(path) unless registered.include?(path)
+    path
+  end
+
+  # shared date to be invariant across all tests in a run:
+  date_static = Hyrax::TimeService.time_in_utc
+  let(:static_date) { date_static }
+
+  # path fixtures:
+  let(:example_gray_jp2) { File.join(fixture_path, 'ocr_gray.jp2') }
+  let(:txt_path) { File.join(fixture_path, 'credits.md') }
+  let(:sample_thumbnail) { File.join(fixture_path, 'thumbnail.jpg') }
+
+  # sample data:
+  let(:sample_text) { 'even in a mythical Age there must be some enigmas' }
+
+  let(:valid_file_set) do
+    file_set = FileSet.new
+    file_set.save!(validate: false)
+    file_set
+  end
+
+  let(:sample_work) do
+    work = MyWork.new
+    work.title = ['Bombadil']
+    work.members.push(valid_file_set)
+    work.save!
+    work
+  end
+
+  # sample objects:
+  let(:work_with_file) do
+    # we need a work with not just a valid (but empty) fileset, but also
+    #   a persisted file, so we use the shared work sample, and expand
+    #   on it with actual file data/metadata.
+    work = sample_work
+    fileset = work.members.first
+    file = Hydra::PCDM::File.create
+    fileset.original_file = file
+    # Set binary content on file via ActiveFedora content= mutator method
+    #   which also makes .size method return valid result for content
+    file.content = File.open(txt_path)
+    # Set some metdata we would expect to otherwise be set upon an upload
+    file.original_name = 'credits.md'
+    file.mime_type = 'text/plain'
+    file.date_modified = static_date
+    file.date_created = static_date
+    # saving fileset also saves file content
+    fileset.save!
+    work
+  end
+
+  def path_factory
+    Hyrax::DerivativePath
+  end
+
+  def work_file_set(work)
+    work.members.detect { |m| m.is_a? FileSet }
+  end
+
+  def text_path(work)
+    path_factory.derivative_path_for_reference(work_file_set(work), 'txt')
+  end
+
+  def jp2_path(work)
+    path_factory.derivative_path_for_reference(work_file_set(work), 'jp2')
+  end
+
+  def thumbnail_path(work)
+    path_factory.derivative_path_for_reference(work_file_set(work), 'thumbnail')
+  end
+
+  def mkdir_derivative(work, name)
+    # make shared path for derivatives to live, Hyrax ususally does this
+    #   for thumbnails, and iiif_print does this in its derivative
+    #   service plugins; here we do same.
+    fsid = work_file_set(work).id
+    path = path_factory.derivative_path_for_reference(fsid, name)
+    dir = File.join(path.split('/')[0..-2])
+    FileUtils.mkdir_p(dir) unless Dir.exist?(dir)
+  end
+
+  def mk_jp2_derivative(work)
+    mkdir_derivative(work, 'jp2')
+    dst_path = jp2_path(work)
+    FileUtils.copy(example_gray_jp2, dst_path)
+    expect(File.exist?(dst_path)).to be true
+  end
+
+  def mk_txt_derivative(work)
+    mkdir_derivative(work, 'txt')
+    dst_path = text_path(work)
+    File.open(dst_path, 'w') { |f| f.write(sample_text) }
+    expect(File.exist?(dst_path)).to be true
+  end
+
+  def mk_thumbnail_derivative(work)
+    mkdir_derivative(work, 'thumbnail')
+    dst_path = thumbnail_path(work)
+    FileUtils.copy(sample_thumbnail, dst_path)
+    expect(File.exist?(dst_path)).to be true
+  end
+end

data/spec/models/iiif_print/derivative_attachment_spec.rb

@@ -0,0 +1,37 @@
+require 'spec_helper'
+
+module IiifPrint
+  RSpec.describe DerivativeAttachment, type: :model do
+    it "requires some columns to be considered complete" do
+      model = described_class.create
+      # attempt save without required data; expect failure
+      expect { model.save! }.to raise_exception(ActiveRecord::RecordInvalid)
+    end
+
+    it "saves when constructed with all field values" do
+      model = described_class.create(
+        fileset_id: 'a1b2c3d4e5',
+        path: '/path/to/somefile',
+        destination_name: 'txt'
+      )
+      # attempt save without required data; expect failure
+      expect { model.save! }.not_to raise_exception
+    end
+
+    it "saves when all fields completely set" do
+      model = described_class.create
+      model.fileset_id = 'someid123'
+      model.path = '/path/to/somefile'
+      model.destination_name = 'txt'
+      expect { model.save! }.not_to raise_exception
+    end
+
+    it "saves when only path, destination_name set" do
+      model = described_class.create
+      model.fileset_id = nil
+      model.path = '/path/to/somefile'
+      model.destination_name = 'txt'
+      expect { model.save! }.not_to raise_exception
+    end
+  end
+end

data/spec/models/iiif_print/ingest_file_relation_spec.rb

@@ -0,0 +1,56 @@
+require 'spec_helper'
+
+module IiifPrint
+  RSpec.describe IngestFileRelation, type: :model do
+    def make_test_records
+      # two unique values
+      described_class.create(
+        file_path: '/some/path/to/this',
+        derivative_path: '/some/path/to/that'
+      )
+      described_class.create(
+        file_path: '/some/path/to/this',
+        derivative_path: '/some/path/to/other_thing'
+      )
+      # a duplicate will save, presumption is that dupes are filtered on query:
+      described_class.create(
+        file_path: '/some/path/to/this',
+        derivative_path: '/some/path/to/other_thing'
+      )
+    end
+
+    it "will not save unless record is complete" do
+      model = described_class.create
+      # attempt save without required data; expect failure
+      expect { model.save! }.to raise_exception(ActiveRecord::RecordInvalid)
+      model2 = described_class.create
+      model2.file_path = '/path/to/sourcefile.tiff'
+      expect { model2.save! }.to raise_exception(ActiveRecord::RecordInvalid)
+      model3 = described_class.create
+      model3.derivative_path = '/path/to/sourcefile.tiff'
+      expect { model3.save! }.to raise_exception(ActiveRecord::RecordInvalid)
+    end
+
+    it "will save sufficiently constructed record" do
+      model = described_class.create(
+        file_path: '/path/to/this',
+        derivative_path: '/path/to/that'
+      )
+      expect { model.save! }.not_to raise_exception
+    end
+
+    it "will save when all fields completely set" do
+      model = described_class.create
+      model.file_path = '/path/to/sourcefile.tiff'
+      model.derivative_path = '/path/to/derived.jp2'
+      expect { model.save! }.not_to raise_exception
+    end
+
+    it "can query derivative paths for primary file" do
+      make_test_records
+      result = described_class.derivatives_for_file('/some/path/to/this')
+      expect(result).to be_an Array
+      expect(result.size).to eq 2
+    end
+  end
+end

data/spec/models/solr_document_spec.rb

@@ -0,0 +1,14 @@
+require 'spec_helper'
+RSpec.describe SolrDocument do
+  let(:solr_doc) { described_class.new(id: 'foo', file_set_ids_ssim: ['bar']) }
+
+  describe 'file_set_ids' do
+    it 'responds to #file_set_ids' do
+      expect(solr_doc).to respond_to(:file_set_ids)
+    end
+
+    it 'returns the correct value' do
+      expect(solr_doc.file_set_ids).to eq(['bar'])
+    end
+  end
+end

data/spec/presenters/iiif_print/iiif_manifest_presenter_behavior_spec.rb

@@ -0,0 +1,19 @@
+require 'spec_helper'
+
+RSpec.describe IiifPrint::IiifManifestPresenterBehavior do
+  let(:attributes) do
+    { "id" => "abc123",
+      "title_tesim" => ['Page the first'],
+      "description_tesim" => ['A book or something'],
+      "creator_tesim" => ['Arthur McAuthor'] }
+  end
+  let(:solr_document) { SolrDocument.new(attributes) }
+  let(:presenter) { Hyrax::IiifManifestPresenter.new(solr_document) }
+  let(:test_request) { ActionDispatch::TestRequest.new({}) }
+
+  describe '#search_service' do
+    it 'returns the correct URL for the IIIF Search service' do
+      expect(presenter.search_service).to include("#{solr_document.id}/iiif_search")
+    end
+  end
+end

data/spec/presenters/iiif_print/iiif_manifest_presenter_factory_behavior_spec.rb

@@ -0,0 +1,49 @@
+require "spec_helper"
+
+RSpec.describe IiifPrint::IiifManifestPresenterBehavior do
+  let(:parent_fs_attributes) do
+    { "id" => "parent_fs123",
+      "title_tesim" => ["My Parent FileSet"],
+      "has_model_ssim" => ["FileSet"] }
+  end
+  let(:child_work_attributes) do
+    { "id" => "child_work123",
+      "title_tesim" => ["My Child Image"],
+      "has_model_ssim" => ["Image"],
+      "file_set_ids_ssim" => ["child_image_fs123"] }
+  end
+  let(:child_fs_attributes) do
+    { "id" => "child_fs123",
+    "title_tesim" => ["My Child FileSet"],
+    "has_model_ssim" => ["FileSet"] }
+  end
+  let(:parent_fs_solr_doc) { SolrDocument.new(parent_fs_attributes) }
+  let(:child_work_solr_doc) { SolrDocument.new(child_work_attributes) }
+  let(:child_fs_solr_doc) { SolrDocument.new(child_fs_attributes) }
+  let(:ids) { [parent_fs_solr_doc.id, child_work_solr_doc.id] }
+  let(:presenter_class) { Hyrax::IiifManifestPresenter }
+
+  subject(:presenter_factory) do
+    Hyrax::IiifManifestPresenter::Factory.new(
+      ids: ids,
+      presenter_class: presenter_class,
+      presenter_args: []
+    )
+  end
+
+  describe "#build" do
+    it "returns an Array of DisplayImagePresenters" do
+      allow_any_instance_of(Hyrax::IiifManifestPresenter::Factory)
+        .to receive(:load_docs).and_return([parent_fs_solr_doc, child_work_solr_doc])
+      allow_any_instance_of(IiifPrint::IiifManifestPresenterFactoryBehavior)
+        .to receive(:load_file_set_docs).and_return([child_fs_solr_doc])
+      allow(child_work_solr_doc).to receive(:hydra_model).and_return(MyWork)
+      allow(Hyrax.config).to receive(:curation_concerns).and_return([MyWork])
+
+      expect(subject.build).to be_an Array
+      expect(subject.build.size).to eq ids.size
+      expect(subject.build.map(&:class).uniq.size).to eq 1
+      expect(subject.build.first.class).to eq Hyrax::IiifManifestPresenter::DisplayImagePresenter
+    end
+  end
+end

data/spec/services/iiif_print/jp2_derivative_service_spec.rb

@@ -0,0 +1,59 @@
+require 'spec_helper'
+RSpec.describe IiifPrint::JP2DerivativeService do
+  let(:valid_file_set) do
+    file_set = FileSet.new
+    file_set.save!(validate: false)
+    file_set
+  end
+
+  let(:fixture_path) do
+    File.join(
+      IiifPrint::GEM_PATH, 'spec', 'fixtures', 'files'
+    )
+  end
+
+  describe "Creates JP2 derivatives" do
+    def source_image(name)
+      File.join(fixture_path, name)
+    end
+
+    def expected_path(file_set)
+      Hyrax::DerivativePath.derivative_path_for_reference(file_set, 'jp2')
+    end
+
+    def metadata_match_checker(source, target)
+      target_meta = IiifPrint::ImageTool.new(target).metadata
+      source_meta = IiifPrint::ImageTool.new(source).metadata
+      expect(target_meta[:content_type]).to eq 'image/jp2'
+      expect(target_meta[:width]).to eq source_meta[:width]
+      expect(target_meta[:height]).to eq source_meta[:height]
+    end
+
+    def makes_jp2(filename)
+      expected = expected_path(valid_file_set)
+      expect(File.exist?(expected)).to be false
+      svc = described_class.new(valid_file_set)
+      source_path = source_image(filename)
+      svc.create_derivatives(source_path)
+      expect(File.exist?(expected)).to be true
+      metadata_match_checker(source_path, expected)
+      svc.cleanup_derivatives
+    end
+
+    it "creates gray JP2 derivative from one-bit source" do
+      makes_jp2('ocr_mono.tiff')
+    end
+
+    it "creates gray JP2 from grayscale source" do
+      makes_jp2('lowres-gray-via-ndnp-sample.tiff')
+    end
+
+    it "creates color JP2 from color source" do
+      makes_jp2('4.1.07.tiff')
+    end
+
+    it "creates JP2 from PDF source, robust to multi-page" do
+      makes_jp2('sample-color-newsletter.pdf')
+    end
+  end
+end

data/spec/services/iiif_print/pdf_derivative_service_spec.rb

@@ -0,0 +1,66 @@
+require 'spec_helper'
+RSpec.describe IiifPrint::PDFDerivativeService do
+  let(:valid_file_set) do
+    file_set = FileSet.new
+    file_set.save!(validate: false)
+    file_set
+  end
+
+  let(:fixture_path) do
+    File.join(
+      IiifPrint::GEM_PATH, 'spec', 'fixtures', 'files'
+    )
+  end
+
+  describe "Creates PDF derivatives" do
+    def source_image(name)
+      File.join(fixture_path, name)
+    end
+
+    def expected_path(file_set)
+      Hyrax::DerivativePath.derivative_path_for_reference(file_set, 'pdf')
+    end
+
+    # given output file name, check DPI is 150
+    def check_dpi(expected)
+      metadata = IiifPrint::ImageTool.new(expected).metadata
+      # get width of pdf in points (via imagemagick), should be 864x == 12in
+      page_width = metadata[:width]
+      expect(page_width).to eq 864
+      # get total width of image in pixels from pdfimages -list, ==> 1800
+      image_width = 1800
+      im_list = `pdfimages -list #{expected}`
+      expect(im_list.lines[-1].split(' ')[3]).to eq image_width.to_s
+      # this combination of page pt width, image px width ==> 150ppi
+      expect(image_width / (page_width / 72.0)).to eq 150.0
+    end
+
+    def makes_pdf(filename)
+      expected = expected_path(valid_file_set)
+      expect(File.exist?(expected)).to be false
+      svc = described_class.new(valid_file_set)
+      svc.create_derivatives(source_image(filename))
+      expect(File.exist?(expected)).to be true
+      metadata = IiifPrint::ImageTool.new(expected).metadata
+      expect(metadata[:content_type]).to eq 'application/pdf'
+      check_dpi(expected)
+      svc.cleanup_derivatives
+    end
+
+    it "creates gray PDF derivative from one-bit source" do
+      makes_pdf('ocr_mono.tiff')
+    end
+
+    it "creates gray PDF from grayscale source" do
+      makes_pdf('lowres-gray-via-ndnp-sample.tiff')
+    end
+
+    it "creates color PDF from color source" do
+      makes_pdf('4.1.07.tiff')
+    end
+
+    it "creates color PDF from color JP2 source" do
+      makes_pdf('4.1.07.jp2')
+    end
+  end
+end