newspaper_works 0.1.0

data/spec/actors/newspaper_works/actors/newspaper_works_upload_actor_spec.rb

@@ -0,0 +1,69 @@
+require 'faraday'
+require 'spec_helper'
+require 'misc_shared'
+
+RSpec.describe NewspaperWorks::Actors::NewspaperWorksUploadActor, :perform_enqueued do
+  include_context 'shared setup'
+
+  let(:issue) { build(:newspaper_issue) }
+  let(:ability) { build(:ability) }
+  let(:uploaded_pdf_file) { create(:uploaded_pdf_file) }
+  let(:uploaded_file_ids) { [uploaded_pdf_file.id] }
+  let(:attributes) { { title: ['foo'], uploaded_files: uploaded_file_ids } }
+  let(:terminator) { Hyrax::Actors::Terminator.new }
+  # environment with uploads:
+  let(:env) { Hyrax::Actors::Environment.new(issue, ability, attributes) }
+  # environment with NO uploads:
+  let(:edit_env) { Hyrax::Actors::Environment.new(issue, ability, {}) }
+  let(:middleware) do
+    stack = ActionDispatch::MiddlewareStack.new.tap do |middleware|
+      middleware.use described_class
+    end
+    stack.build(terminator)
+  end
+
+  let(:uploaded_issue) do
+    middleware.public_send(:create, env)
+    # return work, reloaded, because env.curation_concern will be stale after
+    #   running actor.
+    NewspaperIssue.find(env.curation_concern.id)
+  end
+
+  let(:edited_issue) do
+    middleware.public_send(:update, edit_env)
+    NewspaperIssue.find(edit_env.curation_concern.id)
+  end
+
+  describe "NewspaperIssue upload of PDF" do
+    do_now_jobs = [
+      NewspaperWorks::CreateIssuePagesJob,
+      IngestLocalFileJob,
+      IngestJob
+    ]
+
+    # we over-burden one example, because sadly RSpec does not do well with
+    #   shared state across examples (without use of `before(:all)` which is
+    #   mutually exclusive with `let` in practice, and ruffles rubocop's
+    #   overzealous sense of moral duty, speaking of which:
+    it "creates child pages for issue", perform_enqueued: do_now_jobs do
+      pages = uploaded_issue.ordered_pages
+      expect(pages.size).to eq 2
+      page = pages[0]
+      # Page needs correct admin set:
+      expect(page.admin_set_id).to eq 'admin_set/default'
+      file_sets = page.members.select { |v| v.class == FileSet }
+      expect(file_sets.size).to eq 1
+      files = file_sets[0].files
+      url = files[0].uri.to_s
+      # fetch the thing from Fedora Commons:
+      response = Faraday.get(url)
+      stored_size = response.body.length
+      expect(stored_size).to be > 0
+      # expect that subsequent edits of same issue (run though update
+      #   method of actor stack) do not duplicate pages (verify by count):
+      expect(edited_issue.id).to eq uploaded_issue.id
+      pages = edited_issue.ordered_pages
+      expect(pages.size).to eq 2 # still the same page count
+    end
+  end
+end

data/spec/controllers/catalog_controller_spec.rb

@@ -0,0 +1,63 @@
+require 'spec_helper'
+
+# specs to test that install generators have correctly modified CatalogController
+RSpec.describe CatalogController do
+  describe 'NewspaperWorks::InstallGenerator#add_index_fields_to_catalog_controller' do
+    subject { described_class.blacklight_config.index_fields }
+
+    it 'has NewspaperWorks index fields' do
+      expect(subject['publication_title_ssi']).not_to be_falsey
+      expect(subject['place_of_publication_label_tesim']).not_to be_falsey
+      expect(subject['publication_date_dtsi'].class).to eq(Blacklight::Configuration::IndexField)
+    end
+  end
+
+  describe 'NewspaperWorks::InstallGenerator#add_facets_to_catalog_controller' do
+    subject { described_class.blacklight_config.facet_fields }
+
+    it 'has NewspaperWorks facet fields' do
+      expect(subject['place_of_publication_city_sim']).not_to be_falsey
+      expect(subject['publication_title_ssi'].class).to eq(Blacklight::Configuration::FacetField)
+      expect(subject['genre_sim'].label).to eq('Article type')
+    end
+
+    it 'has definitions for non-displaying facet fields' do
+      expect(subject['place_of_publication_label_sim']).not_to be_falsey
+      expect(subject['issn_sim']).not_to be_falsey
+      expect(subject['lccn_sim']).not_to be_falsey
+      expect(subject['oclcnum_sim']).not_to be_falsey
+      expect(subject['held_by_sim']).not_to be_falsey
+      expect(subject['author_sim']).not_to be_falsey
+      expect(subject['photographer_sim']).not_to be_falsey
+      expect(subject['geographic_coverage_sim']).not_to be_falsey
+      expect(subject['preceded_by_sim']).not_to be_falsey
+      expect(subject['succeeded_by_sim']).not_to be_falsey
+    end
+  end
+
+  describe 'NewspaperWorks::InstallGenerator#add_pubdate_sort_to_catalog_controller' do
+    subject { described_class.blacklight_config.sort_fields }
+
+    it 'has NewspaperWorks sort fields' do
+      expect(subject['publication_date_dtsi asc'].class).to eq Blacklight::Configuration::SortField
+      expect(subject['publication_date_dtsi desc'].field).to eq 'publication_date_dtsi desc'
+    end
+  end
+
+  describe 'NewspaperWorks::BlacklightAdvancedSearchGenerator' do
+    subject { described_class.blacklight_config }
+
+    describe '#update_search_builder' do
+      it 'sets the default SearchBuilder' do
+        expect(subject.search_builder_class).to eq CustomSearchBuilder
+      end
+    end
+
+    describe '#add_newspapers_advanced_config' do
+      it 'adds the advanced_search[:newspapers_search] config' do
+        expect(subject.advanced_search[:newspapers_search]).not_to be_falsey
+        expect(subject.advanced_search[:newspapers_search][:form_solr_parameters]['facet.field']).not_to be_blank
+      end
+    end
+  end
+end

data/spec/controllers/newspaper_works/newspapers_controller_spec.rb

@@ -0,0 +1,114 @@
+require 'spec_helper'
+
+RSpec.describe NewspaperWorks::NewspapersController do
+  pubdate = '2015-03-15'
+
+  # have to build a set of related objects first, can't use model_shared
+  # because we need a unique lccn every time specs are run
+  # otherwise query in NewspaperController#find_object returns false
+  # use before(:all) so we only create fixtures once
+  before(:all) do
+    @lccn = "sn#{rand(10_000)}"
+
+    title = NewspaperTitle.new
+    title.title = ["Yesterday's News"]
+    title.lccn = @lccn
+    title.visibility = 'open'
+
+    issue = NewspaperIssue.new
+    issue.title = ["Yesterday's News: December 7, 1941"]
+    issue.publication_date = pubdate
+    issue.visibility = 'open'
+
+    title.members.push issue
+
+    page1 = NewspaperPage.new
+    page1.title = ['Page 1']
+    page1.visibility = 'open'
+    page2 = NewspaperPage.new
+    page2.title = ['Page 2']
+    page2.visibility = 'open'
+
+    issue.ordered_members << page1
+    issue.ordered_members << page2
+
+    issue.save!
+    title.save!
+    page1.save!
+    page2.save!
+  end
+
+  describe 'GET "title"' do
+    describe 'with a valid unique_id' do
+      it 'renders the page' do
+        get :title, params: { unique_id: @lccn }
+        expect(response).to be_successful
+        expect(response).to render_template('hyrax/newspaper_titles/show')
+      end
+    end
+
+    describe 'with an invalid unique_id' do
+      it 'returns an error' do
+        expect do
+          get :title, params: { unique_id: 'foo' }
+        end.to raise_error ActionController::RoutingError
+      end
+    end
+  end
+
+  describe 'GET "issue"' do
+    describe 'with a valid date' do
+      it 'renders the page' do
+        get :issue, params: { unique_id: @lccn, date: pubdate }
+        expect(response).to be_successful
+        expect(response).to render_template('hyrax/newspaper_issues/show')
+      end
+    end
+
+    describe 'with an edition param' do
+      it 'renders the page' do
+        get :issue, params: { unique_id: @lccn, date: pubdate, edition: 'ed-1' }
+        expect(response).to be_successful
+        expect(response).to render_template('hyrax/newspaper_issues/show')
+      end
+    end
+
+    describe 'with an invalid date' do
+      it 'returns an error' do
+        expect do
+          get :issue, params: { unique_id: @lccn, date: '2015-02-31' }
+        end.to raise_error ActionController::RoutingError
+      end
+    end
+  end
+
+  describe 'GET "page"' do
+    describe 'with valid params' do
+      it 'renders the page' do
+        get :page,
+            params: {
+              unique_id: @lccn,
+              date: pubdate,
+              edition: 'ed-1',
+              page: 'seq-1'
+            }
+        expect(response).to be_successful
+        expect(response).to render_template('hyrax/newspaper_pages/show')
+      end
+    end
+
+    describe 'with invalid page params' do
+      it 'returns an error' do
+        expect do
+          get :page,
+              params: {
+                unique_id: @lccn,
+                date: pubdate,
+                edition: 'ed-1',
+                page: 'seq-3'
+              }
+        end.to raise_error ActionController::RoutingError
+      end
+    end
+  end
+end

data/spec/controllers/newspaper_works/newspapers_search_controller_spec.rb

@@ -0,0 +1,21 @@
+require 'spec_helper'
+
+RSpec.describe NewspaperWorks::NewspapersSearchController do
+  describe 'GET "search"' do
+    before { get :search }
+
+    it 'renders the page' do
+      expect(response).to be_successful
+      expect(response).to render_template('newspaper_works/newspapers_search/search')
+    end
+
+    it 'sets the @response' do
+      expect(assigns(:response)).to be_a_kind_of(Blacklight::Solr::Response)
+    end
+  end
+
+  describe '#search_builder_class' do
+    subject { controller.search_builder_class }
+    it { is_expected.to eq NewspaperWorks::NewspapersSearchBuilder }
+  end
+end

data/spec/factories/ability.rb

@@ -0,0 +1,6 @@
+FactoryBot.define do
+  factory :ability do
+    user
+    initialize_with { new(user) }
+  end
+end

data/spec/factories/newspaper_issue.rb

@@ -0,0 +1,7 @@
+# will infer, create a NewspaperIssue object
+FactoryBot.define do
+  factory :newspaper_issue do
+    title { ['Here and There'] }
+    depositor { User.batch_user.user_key }
+  end
+end

data/spec/factories/newspaper_issue_ingest.rb

@@ -0,0 +1,6 @@
+FactoryBot.define do
+  factory :newspaper_issue_ingest, class: NewspaperWorks::Ingest::NewspaperIssueIngest do
+    newspaper_issue
+    initialize_with { new(newspaper_issue) }
+  end
+end

data/spec/factories/newspaper_page.rb

@@ -0,0 +1,7 @@
+# will infer, create a NewspaperPage object
+FactoryBot.define do
+  factory :newspaper_page do
+    title { ['Here and There'] }
+    depositor { User.batch_user.user_key }
+  end
+end

data/spec/factories/newspaper_page_ingest.rb

@@ -0,0 +1,6 @@
+FactoryBot.define do
+  factory :newspaper_page_ingest, class: NewspaperWorks::Ingest::NewspaperPageIngest do
+    newspaper_page
+    initialize_with { new(newspaper_page) }
+  end
+end

data/spec/factories/newspaper_page_solr_document.rb

@@ -0,0 +1,12 @@
+FactoryBot.define do
+  factory :newspaper_page_solr_document, class: SolrDocument do
+    initialize_with do
+      new(id: '123456',
+          title_tesim: ['Page 1'],
+          has_model_ssim: ['NewspaperPage'],
+          issue_id_ssi: 'abc123',
+          file_set_ids_ssim: ['7891011'],
+          thumbnail_path_ss: '/downloads/123456?file=thumbnail')
+    end
+  end
+end

data/spec/factories/newspaper_title.rb

@@ -0,0 +1,8 @@
+# will infer, create a NewspaperTitle object
+FactoryBot.define do
+  factory :newspaper_title do
+    title { ['ACME Press'] }
+    lccn { 'sn2036999999' }
+    depositor { User.batch_user.user_key }
+  end
+end

data/spec/factories/uploaded_pdf_file.rb

@@ -0,0 +1,9 @@
+FactoryBot.define do
+  factory :uploaded_pdf_file, class: Hyrax::UploadedFile do
+    initialize_with do
+      base = File.join(NewspaperWorks::GEM_PATH, 'spec', 'fixtures', 'files')
+      pdf_path = File.join(base, 'minimal-2-page.pdf')
+      new(file: File.open(pdf_path), user: create(:user))
+    end
+  end
+end

data/spec/factories/user.rb

@@ -0,0 +1,13 @@
+FactoryBot.define do
+  factory :user do
+    id { "skroob" }
+    email { "spaceballs@example.com" }
+    password { "password_is_12345" }
+    initialize_with do
+      User.find_or_create_by(id: id) do |user|
+        user.email = email
+        user.password = password
+      end
+    end
+  end
+end

data/spec/features/front_pages_for_title_spec.rb

@@ -0,0 +1,19 @@
+require 'spec_helper'
+require 'features_shared'
+
+RSpec.describe 'front_pages_for_title' do
+  include_context "fixtures_for_features"
+
+  # @title1 comes from fixtures_for_features
+  before do
+    visit hyrax_newspaper_title_path(@title1.id)
+    click_link('front_pages_search')
+  end
+
+  it 'returns the front pages' do
+    within "#search-results" do
+      expect(page).to have_content "#{title1_issue1_title_memo}: Page 1"
+      expect(page).not_to have_content "#{title1_issue1_title_memo}: Page 2"
+    end
+  end
+end

data/spec/features/newspaper_title_search_spec.rb

@@ -0,0 +1,30 @@
+require 'spec_helper'
+require 'features_shared'
+
+RSpec.describe 'newspaper title search' do
+  include_context "fixtures_for_features"
+
+  # @title1 comes from fixtures_for_features
+  before do
+    visit hyrax_newspaper_title_path(@title1.id)
+    fill_in "q_issues", with: title_base_memo
+  end
+
+  it 'returns pages for this title' do
+    click_button('issue-search')
+    within "#search-results" do
+      expect(page).to have_content "#{title1_issue1_title_memo}: Page 1"
+      expect(page).to have_content "#{title1_issue1_title_memo}: Page 2"
+      expect(page).not_to have_content "#{title2_issue1_title_memo}: Page 1"
+    end
+  end
+
+  it 'returns only front pages if checked' do
+    check 'f_first_page_bsi_'
+    click_button('issue-search')
+    within "#search-results" do
+      expect(page).to have_content "#{title1_issue1_title_memo}: Page 1"
+      expect(page).not_to have_content "#{title1_issue1_title_memo}: Page 2"
+    end
+  end
+end

data/spec/features/newspapers_search_spec.rb

@@ -0,0 +1,49 @@
+require 'spec_helper'
+require 'features_shared'
+
+RSpec.describe 'newspapers_search' do
+  include_context "fixtures_for_features"
+
+  # title_base_memo comes from fixtures_for_features
+  before do
+    visit newspaper_works.newspapers_search_path
+    fill_in "all_fields", with: title_base_memo
+  end
+
+  it 'returns results for keyword search' do
+    click_button('search-submit-newspapers')
+    within "#search-results" do
+      expect(page).to have_content "#{title1_issue1_title_memo}: Page 1"
+      expect(page).to have_content "#{title1_issue1_title_memo}: Page 2"
+      expect(page).to have_content "#{title2_issue1_title_memo}: Page 1"
+    end
+  end
+
+  it 'returns correct results for keyword search with front page' do
+    check 'f_first_page_bsi_'
+    click_button('search-submit-newspapers')
+    within "#search-results" do
+      expect(page).to have_content "#{title2_issue1_title_memo}: Page 1"
+      expect(page).not_to have_content "#{title1_issue1_title_memo}: Page 2"
+    end
+  end
+
+  it 'returns correct results for keyword search with date' do
+    fill_in "date_range_start", with: '1965'
+    fill_in "date_range_end", with: '1966'
+    click_button('search-submit-newspapers')
+    within "#search-results" do
+      expect(page).to have_content "#{title1_issue1_title_memo}: Page 1"
+      expect(page).not_to have_content "#{title2_issue1_title_memo}: Page 1"
+    end
+  end
+
+  it 'returns correct results for keyword search with facet' do
+    check 'f_inclusive_language_sim_spanish'
+    click_button('search-submit-newspapers')
+    within "#search-results" do
+      expect(page).to have_content "#{title2_issue1_title_memo}: Page 1"
+      expect(page).not_to have_content "#{title1_issue1_title_memo}: Page 1"
+    end
+  end
+end

data/spec/features/search_results_thumbnail_highlights_spec.rb

@@ -0,0 +1,33 @@
+require 'spec_helper'
+
+RSpec.describe 'thumbnail_highlights', js: true do
+  fixture_path = File.join(NewspaperWorks::GEM_PATH, 'spec', 'fixtures', 'files')
+  let(:query_term) { 'rotunda' }
+
+  # use before(:all) so we only create fixtures once
+  # we use instance var for @work so we can access its id in specs
+  before(:all) do
+    whitelist = Hyrax.config.whitelisted_ingest_dirs
+    whitelist.push(fixture_path) unless whitelist.include?(fixture_path)
+
+    @work = NewspaperPage.create!(
+      title: ['Test Page for Thumbnail Highlights'],
+      visibility: "open"
+    )
+    attachment = NewspaperWorks::Data::WorkFiles.of(@work)
+    attachment.assign(File.join(fixture_path, 'page1.tiff'))
+    attachment.derivatives.assign(File.join(fixture_path, 'ndnp-sample1-txt.txt'))
+    attachment.derivatives.assign(File.join(fixture_path, 'ndnp-sample1-json.json'))
+    attachment.commit!
+    @work.save!
+  end
+
+  describe 'thumbnail highlighting' do
+    it 'adds highlight divs to thumbnail' do
+      visit search_catalog_path(q: 'rotunda scarcely')
+      within "#document_#{@work.id}" do
+        expect(page).to have_selector('.thumbnail_highlight', count: 2)
+      end
+    end
+  end
+end

data/spec/features_shared.rb

@@ -0,0 +1,71 @@
+# a set of Title, Issue, and Page objects that can be reused in feature specs
+RSpec.shared_context "fixtures_for_features", shared_context: :metadata do
+  # use this so titles are different every time spec is run
+  # prevents tests from accidentally failing because search results view only shows 10 results
+  # and our newly created fixtures may not show up on first results page
+  title_base = 'Mxyzptlk'.chars.shuffle.join
+  title1_title = "#{title_base} Examiner"
+  title1_issue1_title = "#{title1_title}: July 4, 1965"
+  title2_title = "#{title_base} Courier"
+  title2_issue1_title = "#{title2_title}: July 4, 1969"
+
+  # memo-ize these so we can use it in specs
+  let!(:title_base_memo) { title_base }
+  let!(:title1_issue1_title_memo) { title1_issue1_title }
+  let!(:title2_issue1_title_memo) { title2_issue1_title }
+
+  # we use instance var for @title1 so we can access its id in specs
+  # # use before(:all) so we only create fixtures once
+  before(:all) do
+    @title1 = NewspaperTitle.new
+    @title1.title = [title1_title]
+    @title1.lccn = 'sn0000000'
+    @title1.visibility = 'open'
+
+    title1_issue1 = NewspaperIssue.new
+    title1_issue1.title = [title1_issue1_title]
+    title1_issue1.resource_type = ["newspaper"]
+    title1_issue1.language = ["English"]
+    title1_issue1.held_by = "Marriott Library"
+    title1_issue1.publication_date = '1965-07-04'
+    title1_issue1.visibility = 'open'
+    @title1.members << title1_issue1
+
+    title1_issue1_page1 = NewspaperPage.new
+    title1_issue1_page1.title = ["#{title1_issue1_title}: Page 1"]
+    title1_issue1_page1.visibility = 'open'
+    title1_issue1_page2 = NewspaperPage.new
+    title1_issue1_page2.title = ["#{title1_issue1_title}: Page 2"]
+    title1_issue1_page2.visibility = 'open'
+    title1_issue1.ordered_members << title1_issue1_page1
+    title1_issue1.ordered_members << title1_issue1_page2
+
+    title1_issue1.save
+    @title1.save
+    title1_issue1_page1.save
+    title1_issue1_page2.save
+
+    title2 = NewspaperTitle.new
+    title2.title = [title2_title]
+    title2.lccn = 'sn1111111'
+    title2.visibility = 'open'
+
+    title2_issue1 = NewspaperIssue.new
+    title2_issue1.title = [title2_issue1_title]
+    title2_issue1.resource_type = ["newspaper"]
+    title2_issue1.language = ["Spanish"]
+    title2_issue1.held_by = "Boston Public Library"
+    title2_issue1.publication_date = '1969-07-04'
+    title2_issue1.visibility = 'open'
+    title2.members << title2_issue1
+
+    title2_issue1_page1 = NewspaperPage.new
+    title2_issue1_page1.title = ["#{title2_issue1_title}: Page 1"]
+    title2_issue1_page1.visibility = 'open'
+    title2_issue1.ordered_members << title2_issue1_page1
+
+    title2_issue1.save
+    title2.save
+    title2_issue1_page1.save
+  end
+end

data/spec/fixtures/files/README.md

@@ -0,0 +1,7 @@
+## Sample / fixture file manifest
+
+* sample-4page-issue.pdf
+  - original: chicopee-weekly-journal_1856-05-24-p1.pdf
+* page1.tiff
+  - first page from sample-4page-issue.pdf, converted to CCITT Group 4
+    encoded (lossless compressed) TIFF via Ghostscript.