iiif_print 1.1.0 → 2.0.1
Sign up to get free protection for your applications and to get access to all the features.
- checksums.yaml +4 -4
- data/.rubocop.yml +1 -1
- data/Gemfile.lock +2 -2
- data/README.md +4 -0
- data/app/actors/iiif_print/actors/file_set_actor_decorator.rb +1 -1
- data/app/indexers/concerns/iiif_print/child_work_indexer.rb +27 -0
- data/app/indexers/concerns/iiif_print/file_set_indexer.rb +37 -22
- data/{lib → app/jobs}/iiif_print/jobs/application_job.rb +2 -1
- data/{lib → app/jobs}/iiif_print/jobs/child_works_from_pdf_job.rb +14 -9
- data/{lib → app/jobs}/iiif_print/jobs/create_relationships_job.rb +10 -20
- data/app/listeners/iiif_print/listener.rb +31 -0
- data/app/models/concerns/iiif_print/set_child_flag.rb +1 -1
- data/app/models/concerns/iiif_print/solr/document.rb +5 -3
- data/app/presenters/iiif_print/file_set_presenter_decorator.rb +11 -0
- data/app/presenters/iiif_print/iiif_manifest_presenter_factory_behavior.rb +1 -1
- data/app/presenters/iiif_print/work_show_presenter_decorator.rb +5 -2
- data/app/services/iiif_print/manifest_builder_service_behavior.rb +4 -2
- data/app/services/iiif_print/pluggable_derivative_service.rb +5 -1
- data/app/services/iiif_print/simple_schema_loader_decorator.rb +11 -0
- data/app/transactions/hyrax/transactions/iiif_print_container_decorator.rb +34 -0
- data/app/transactions/hyrax/transactions/steps/conditionally_destroy_children_from_split.rb +32 -0
- data/app/transactions/hyrax/transactions/steps/delete_all_file_sets_decorator.rb +35 -0
- data/app/views/hyrax/file_sets/_show_actions.html.erb +1 -1
- data/config/initializers/simple_schema_loader.rb +1 -0
- data/config/metadata/child_works_from_pdf_splitting.yaml +17 -0
- data/db/migrate/20181214181358_create_iiif_print_derivative_attachments.rb +8 -6
- data/db/migrate/20190107165909_create_iiif_print_ingest_file_relations.rb +7 -5
- data/db/migrate/20230109000000_create_iiif_print_pending_relationships.rb +8 -6
- data/db/migrate/20231110163052_add_model_details_to_iiif_print_pending_relationships.rb +3 -3
- data/iiif_print.gemspec +1 -1
- data/lib/iiif_print/base_derivative_service.rb +13 -2
- data/lib/iiif_print/blacklight_iiif_search/annotation_decorator.rb +2 -2
- data/lib/iiif_print/catalog_search_builder.rb +2 -2
- data/lib/iiif_print/configuration.rb +65 -5
- data/lib/iiif_print/data/fileset_helper.rb +2 -2
- data/lib/iiif_print/data/work_derivatives.rb +1 -1
- data/lib/iiif_print/engine.rb +46 -2
- data/lib/iiif_print/homepage_search_builder.rb +2 -2
- data/lib/iiif_print/jp2_derivative_service.rb +4 -1
- data/lib/iiif_print/lineage_service.rb +19 -6
- data/lib/iiif_print/pdf_derivative_service.rb +3 -1
- data/lib/iiif_print/persistence_layer/active_fedora_adapter.rb +189 -0
- data/lib/iiif_print/persistence_layer/valkyrie_adapter.rb +183 -0
- data/lib/iiif_print/persistence_layer.rb +118 -0
- data/lib/iiif_print/split_pdfs/base_splitter.rb +11 -0
- data/lib/iiif_print/split_pdfs/child_work_creation_from_pdf_service.rb +19 -9
- data/lib/iiif_print/split_pdfs/destroy_pdf_child_works_service.rb +5 -16
- data/lib/iiif_print/text_extraction_derivative_service.rb +4 -2
- data/lib/iiif_print/text_formats_from_alto_service.rb +3 -1
- data/lib/iiif_print/tiff_derivative_service.rb +3 -1
- data/lib/iiif_print/version.rb +1 -1
- data/lib/iiif_print.rb +79 -44
- metadata +19 -192
- data/app/indexers/concerns/iiif_print/child_indexer.rb +0 -40
- data/app/views/hyrax/file_sets/_actions.html.erb +0 -46
- data/bin/rails +0 -13
- data/spec/.keep.txt +0 -1
- data/spec/factories/ability.rb +0 -6
- data/spec/factories/newspaper_issue.rb +0 -7
- data/spec/factories/newspaper_page.rb +0 -7
- data/spec/factories/newspaper_page_solr_document.rb +0 -20
- data/spec/factories/newspaper_title.rb +0 -8
- data/spec/factories/uploaded_pdf_file.rb +0 -9
- data/spec/factories/uploaded_txt_file.rb +0 -9
- data/spec/factories/user.rb +0 -13
- data/spec/fixtures/authorities/licenses.yml +0 -4
- data/spec/fixtures/authorities/rights_statements.yml +0 -4
- data/spec/fixtures/files/4.1.07.jp2 +0 -0
- data/spec/fixtures/files/4.1.07.tiff +0 -0
- data/spec/fixtures/files/README.md +0 -7
- data/spec/fixtures/files/alto-2-0.xsd +0 -714
- data/spec/fixtures/files/broken-truncated.pdf +0 -0
- data/spec/fixtures/files/credits.md +0 -16
- data/spec/fixtures/files/lowres-gray-via-ndnp-sample.tiff +0 -0
- data/spec/fixtures/files/minimal-1-page.pdf +0 -0
- data/spec/fixtures/files/minimal-2-page.pdf +0 -0
- data/spec/fixtures/files/minimal-alto.xml +0 -31
- data/spec/fixtures/files/ndnp-alto-sample.xml +0 -24
- data/spec/fixtures/files/ndnp-sample1-json.json +0 -1
- data/spec/fixtures/files/ndnp-sample1-txt.txt +0 -1
- data/spec/fixtures/files/ndnp-sample1.pdf +0 -0
- data/spec/fixtures/files/ocr_alto.xml +0 -202
- data/spec/fixtures/files/ocr_alto_scaled_4pts_per_px.xml +0 -202
- data/spec/fixtures/files/ocr_color.tiff +0 -0
- data/spec/fixtures/files/ocr_gray.jp2 +0 -0
- data/spec/fixtures/files/ocr_gray.tiff +0 -0
- data/spec/fixtures/files/ocr_mono.tiff +0 -0
- data/spec/fixtures/files/ocr_mono_text_hocr.html +0 -78
- data/spec/fixtures/files/page1.tiff +0 -0
- data/spec/fixtures/files/sample-4page-issue.pdf +0 -0
- data/spec/fixtures/files/sample-color-newsletter.pdf +0 -0
- data/spec/fixtures/files/thumbnail.jpg +0 -0
- data/spec/helpers/hyrax/iiif_helper_spec.rb +0 -65
- data/spec/helpers/iiif_print_helper_spec.rb +0 -43
- data/spec/iiif_print/base_derivative_service_spec.rb +0 -28
- data/spec/iiif_print/blacklight_iiif_search/annotation_decorator_spec.rb +0 -59
- data/spec/iiif_print/catalog_search_builder_spec.rb +0 -60
- data/spec/iiif_print/configuration_spec.rb +0 -193
- data/spec/iiif_print/data/work_derivatives_spec.rb +0 -245
- data/spec/iiif_print/data/work_file_spec.rb +0 -99
- data/spec/iiif_print/data/work_files_spec.rb +0 -237
- data/spec/iiif_print/image_tool_spec.rb +0 -109
- data/spec/iiif_print/jobs/child_works_from_pdf_job_spec.rb +0 -35
- data/spec/iiif_print/jobs/create_relationships_job_spec.rb +0 -118
- data/spec/iiif_print/jp2_image_metadata_spec.rb +0 -37
- data/spec/iiif_print/lineage_service_spec.rb +0 -13
- data/spec/iiif_print/metadata_spec.rb +0 -249
- data/spec/iiif_print/split_pdfs/base_splitter_spec.rb +0 -27
- data/spec/iiif_print/split_pdfs/derivative_rodeo_splitter_spec.rb +0 -80
- data/spec/iiif_print/split_pdfs/destroy_pdf_child_works_service_spec.rb +0 -92
- data/spec/iiif_print/split_pdfs/pages_to_jpgs_splitter_spec.rb +0 -22
- data/spec/iiif_print/split_pdfs/pages_to_pngs_splitter_spec.rb +0 -18
- data/spec/iiif_print/split_pdfs/pages_to_tiffs_splitter_spec.rb +0 -19
- data/spec/iiif_print/text_extraction/alto_reader_spec.rb +0 -49
- data/spec/iiif_print/text_extraction/hocr_reader_spec.rb +0 -45
- data/spec/iiif_print/text_extraction/page_ocr_spec.rb +0 -84
- data/spec/iiif_print/text_extraction/render_alto_spec.rb +0 -54
- data/spec/iiif_print/text_extraction/word_coords_builder_spec.rb +0 -44
- data/spec/iiif_print_spec.rb +0 -171
- data/spec/misc_shared.rb +0 -111
- data/spec/models/iiif_print/derivative_attachment_spec.rb +0 -37
- data/spec/models/iiif_print/iiif_search_decorator_spec.rb +0 -27
- data/spec/models/iiif_print/ingest_file_relation_spec.rb +0 -56
- data/spec/models/solr_document_spec.rb +0 -14
- data/spec/presenters/iiif_print/iiif_manifest_presenter_behavior_spec.rb +0 -70
- data/spec/presenters/iiif_print/iiif_manifest_presenter_factory_behavior_spec.rb +0 -49
- data/spec/samvera/derivatives/configuration_spec.rb +0 -41
- data/spec/samvera/derivatives/hyrax_spec.rb +0 -62
- data/spec/samvera/derivatives_spec.rb +0 -54
- data/spec/services/iiif_print/derivative_rodeo_service_spec.rb +0 -103
- data/spec/services/iiif_print/jp2_derivative_service_spec.rb +0 -59
- data/spec/services/iiif_print/manifest_builder_service_behavior_spec.rb +0 -20
- data/spec/services/iiif_print/pdf_derivative_service_spec.rb +0 -66
- data/spec/services/iiif_print/pluggable_derivative_service_spec.rb +0 -175
- data/spec/services/iiif_print/text_extraction_derivative_service_spec.rb +0 -82
- data/spec/services/iiif_print/text_formats_from_alto_service_spec.rb +0 -127
- data/spec/services/iiif_print/tiff_derivative_service_spec.rb +0 -65
- data/spec/spec_helper.rb +0 -181
- data/spec/support/controller_level_helpers.rb +0 -28
- data/spec/support/iiif_print_models.rb +0 -127
- data/spec/test_app_templates/blacklight.yml +0 -9
- data/spec/test_app_templates/fedora.yml +0 -15
- data/spec/test_app_templates/lib/generators/test_app_generator.rb +0 -40
- data/spec/test_app_templates/redis.yml +0 -9
- data/spec/test_app_templates/solr/conf/schema.xml +0 -362
- data/spec/test_app_templates/solr/conf/solrconfig.xml +0 -322
- data/spec/test_app_templates/solr.yml +0 -7
- /data/{lib → app/jobs}/iiif_print/jobs/request_split_pdf_job.rb +0 -0
@@ -1,65 +0,0 @@
|
|
1
|
-
# frozen_string_literal: true
|
2
|
-
require "spec_helper"
|
3
|
-
|
4
|
-
# @note This spec was brought over from Hyrax 3.4.2
|
5
|
-
# @see https://github.com/samvera/hyrax/blob/afdda8240494ed382301f7d0ab0fd7bafe79185e/spec/helpers/hyrax/iiif_helper_spec.rb
|
6
|
-
RSpec.describe Hyrax::IiifHelper, type: :helper do
|
7
|
-
let(:solr_document) { SolrDocument.new }
|
8
|
-
let(:request) { double }
|
9
|
-
let(:ability) { nil }
|
10
|
-
let(:presenter) { Hyrax::WorkShowPresenter.new(solr_document, ability, request) }
|
11
|
-
let(:uv_partial_path) { 'hyrax/base/iiif_viewers/universal_viewer' }
|
12
|
-
|
13
|
-
describe '#iiif_viewer_display' do
|
14
|
-
before do
|
15
|
-
allow(helper).to receive(:iiif_viewer_display_partial).with(presenter)
|
16
|
-
.and_return(uv_partial_path)
|
17
|
-
end
|
18
|
-
|
19
|
-
it "renders a partial" do
|
20
|
-
expect(helper).to receive(:render)
|
21
|
-
.with(uv_partial_path, presenter: presenter)
|
22
|
-
helper.iiif_viewer_display(presenter)
|
23
|
-
end
|
24
|
-
|
25
|
-
it "takes options" do
|
26
|
-
expect(helper).to receive(:render)
|
27
|
-
.with(uv_partial_path, presenter: presenter, transcript_id: '123')
|
28
|
-
helper.iiif_viewer_display(presenter, transcript_id: '123')
|
29
|
-
end
|
30
|
-
end
|
31
|
-
|
32
|
-
describe '#iiif_viewer_display_partial' do
|
33
|
-
subject { helper.iiif_viewer_display_partial(presenter) }
|
34
|
-
|
35
|
-
it 'defaults to universal viewer' do
|
36
|
-
expect(subject).to eq uv_partial_path
|
37
|
-
end
|
38
|
-
|
39
|
-
context "with #iiif_viewer override" do
|
40
|
-
let(:iiif_viewer) { :mirador }
|
41
|
-
|
42
|
-
before do
|
43
|
-
allow(presenter).to receive(:iiif_viewer).and_return(iiif_viewer)
|
44
|
-
end
|
45
|
-
|
46
|
-
it { is_expected.to eq 'hyrax/base/iiif_viewers/mirador' }
|
47
|
-
end
|
48
|
-
end
|
49
|
-
|
50
|
-
describe '#universal_viewer_base_url' do
|
51
|
-
subject { helper.universal_viewer_base_url }
|
52
|
-
|
53
|
-
it 'defaults to universal viewer base path' do
|
54
|
-
expect(subject).to eq "http://test.host/uv/uv.html"
|
55
|
-
end
|
56
|
-
end
|
57
|
-
|
58
|
-
describe '#universal_viewer_config_url' do
|
59
|
-
subject { helper.universal_viewer_config_url }
|
60
|
-
|
61
|
-
it 'defaults to universal viewer base path' do
|
62
|
-
expect(subject).to eq "http://test.host/uv/uv-config.json"
|
63
|
-
end
|
64
|
-
end
|
65
|
-
end
|
@@ -1,43 +0,0 @@
|
|
1
|
-
require 'spec_helper'
|
2
|
-
|
3
|
-
RSpec.describe IiifPrintHelper do
|
4
|
-
let(:query_term) { 'suffrage' }
|
5
|
-
let(:query_params_hash) { { q: query_term } }
|
6
|
-
let(:document) { build(:newspaper_page_solr_document) }
|
7
|
-
|
8
|
-
describe '#iiif_search_anchor' do
|
9
|
-
it 'returns the correct string' do
|
10
|
-
expect(helper.iiif_search_anchor({})).to eq nil
|
11
|
-
expect(helper.iiif_search_anchor(query_params_hash)).to eq("?h=#{query_term}")
|
12
|
-
end
|
13
|
-
end
|
14
|
-
|
15
|
-
describe '#search_query' do
|
16
|
-
it 'returns the correct string' do
|
17
|
-
expect(helper.search_query({})).to eq nil
|
18
|
-
expect(helper.search_query(query_params_hash)).to eq(query_term)
|
19
|
-
end
|
20
|
-
end
|
21
|
-
|
22
|
-
describe '#highlight_matches' do
|
23
|
-
let(:hl_fl) { 'all_text_tsimv' }
|
24
|
-
|
25
|
-
describe 'when highlighting is present in Solr response' do
|
26
|
-
before do
|
27
|
-
allow(document).to receive(:highlight_field).with(hl_fl).and_return(['foo <em>bar</em> baz'.html_safe])
|
28
|
-
end
|
29
|
-
it 'returns the matching terms when highlighting present' do
|
30
|
-
expect(helper.highlight_matches(document, hl_fl, 'em')).to eq 'bar'
|
31
|
-
end
|
32
|
-
end
|
33
|
-
|
34
|
-
describe 'when highlighting is not present' do
|
35
|
-
before do
|
36
|
-
allow(document).to receive(:highlight_field).with(hl_fl).and_return([])
|
37
|
-
end
|
38
|
-
it 'returns the matching terms when highlighting present' do
|
39
|
-
expect(helper.highlight_matches(document, hl_fl, 'em')).to eq nil
|
40
|
-
end
|
41
|
-
end
|
42
|
-
end
|
43
|
-
end
|
@@ -1,28 +0,0 @@
|
|
1
|
-
require 'spec_helper'
|
2
|
-
|
3
|
-
RSpec.describe IiifPrint::BaseDerivativeService do
|
4
|
-
let(:file_set) { double(FileSet) }
|
5
|
-
let(:service) { described_class.new(file_set) }
|
6
|
-
|
7
|
-
describe '#valid?' do
|
8
|
-
subject { service.valid? }
|
9
|
-
|
10
|
-
context 'when given an image file' do
|
11
|
-
let(:file_set) { double(FileSet, mime_type: 'image/tiff', class: FileSet) }
|
12
|
-
|
13
|
-
it { is_expected.to be_truthy }
|
14
|
-
end
|
15
|
-
|
16
|
-
context 'when given a non-image file' do
|
17
|
-
let(:file_set) { double(FileSet, mime_type: 'audio/mpeg', class: FileSet) }
|
18
|
-
|
19
|
-
it { is_expected.to be_falsey }
|
20
|
-
end
|
21
|
-
end
|
22
|
-
|
23
|
-
describe "instance" do
|
24
|
-
subject { service }
|
25
|
-
|
26
|
-
it { is_expected.to respond_to :target_extension }
|
27
|
-
end
|
28
|
-
end
|
@@ -1,59 +0,0 @@
|
|
1
|
-
require 'spec_helper'
|
2
|
-
|
3
|
-
RSpec.describe IiifPrint::BlacklightIiifSearch::AnnotationDecorator do
|
4
|
-
let(:parent_id) { 'abc123' }
|
5
|
-
let(:page_document) do
|
6
|
-
doc = build(:newspaper_page_solr_document)
|
7
|
-
doc['is_page_of_ssim'] = [parent_id]
|
8
|
-
doc
|
9
|
-
end
|
10
|
-
let(:controller) { CatalogController.new }
|
11
|
-
let(:coordinates) do
|
12
|
-
JSON.parse("{\"coords\":{\"software\":[[2641,4102,512,44]]}}")
|
13
|
-
end
|
14
|
-
let(:parent_document) do
|
15
|
-
SolrDocument.new('id' => parent_id,
|
16
|
-
'has_model_ssim' => ['NewspaperIssue'])
|
17
|
-
end
|
18
|
-
let(:query) { "software AND (is_page_of_ssim:#{parent_id} OR id:#{parent_id})" }
|
19
|
-
let(:iiif_search_annotation) do
|
20
|
-
BlacklightIiifSearch::IiifSearchAnnotation.new(page_document, query,
|
21
|
-
0, nil, controller,
|
22
|
-
parent_document)
|
23
|
-
end
|
24
|
-
let(:file_set) { build(:file_set_solr_document) }
|
25
|
-
let(:test_request) { ActionDispatch::TestRequest.new({}) }
|
26
|
-
|
27
|
-
before do
|
28
|
-
allow(controller).to receive(:request).and_return(test_request)
|
29
|
-
allow(controller).to receive(:polymorphic_url)
|
30
|
-
.with(parent_document, host: test_request.base_url, locale: nil)
|
31
|
-
.and_return("/#{page_document[:issue_id_ssi]}")
|
32
|
-
allow(SolrDocument).to receive(:find).with(file_set.id).and_return(file_set)
|
33
|
-
end
|
34
|
-
|
35
|
-
describe '#annotation_id' do
|
36
|
-
subject { iiif_search_annotation.annotation_id }
|
37
|
-
it 'returns a properly formatted URL' do
|
38
|
-
expect(subject).to include("#{page_document[:issue_id_ssi]}/manifest/canvas/#{page_document[:file_set_ids_ssim].first}/annotation/0")
|
39
|
-
end
|
40
|
-
end
|
41
|
-
|
42
|
-
describe '#canvas_uri_for_annotation' do
|
43
|
-
before { allow(iiif_search_annotation).to receive(:fetch_and_parse_coords).and_return(coordinates) }
|
44
|
-
|
45
|
-
subject { iiif_search_annotation.canvas_uri_for_annotation }
|
46
|
-
it 'returns a properly formatted URL' do
|
47
|
-
expect(subject).to include("#{page_document[:issue_id_ssi]}/manifest/canvas/#{page_document[:file_set_ids_ssim].first}")
|
48
|
-
end
|
49
|
-
|
50
|
-
describe 'private methods' do
|
51
|
-
# test #coordinates based on output of #canvas_uri_for_annotation, which calls it
|
52
|
-
describe '#coordinates' do
|
53
|
-
it 'gets the expected value from #coordinates' do
|
54
|
-
expect(subject).to include("#xywh=2641,4102,512,44")
|
55
|
-
end
|
56
|
-
end
|
57
|
-
end
|
58
|
-
end
|
59
|
-
end
|
@@ -1,60 +0,0 @@
|
|
1
|
-
require 'spec_helper'
|
2
|
-
|
3
|
-
RSpec.describe IiifPrint::CatalogSearchBuilder do
|
4
|
-
# specs for IiifPrint::HighlightSearchParams
|
5
|
-
describe 'highlight_search_params' do
|
6
|
-
let(:solr_parameters) { { q: 'abolition' } }
|
7
|
-
subject { described_class.new(solr_parameters) }
|
8
|
-
|
9
|
-
it 'is included in the default_processor_chain' do
|
10
|
-
expect(described_class.default_processor_chain).to include(:highlight_search_params)
|
11
|
-
end
|
12
|
-
|
13
|
-
before { subject.highlight_search_params(solr_parameters) }
|
14
|
-
it 'adds the highlight fields to solr_parameters' do
|
15
|
-
expect(solr_parameters[:hl]).to be_truthy
|
16
|
-
expect(solr_parameters[:'hl.fl']).to eq('*')
|
17
|
-
end
|
18
|
-
end
|
19
|
-
|
20
|
-
# specs for IiifPrint::ExcludeModels
|
21
|
-
describe 'exclude_models' do
|
22
|
-
let(:solr_parameters) { { all_fields: 'prohibition' } }
|
23
|
-
subject { described_class.new(solr_parameters) }
|
24
|
-
|
25
|
-
it 'is included in the default_processor_chain' do
|
26
|
-
expect(described_class.default_processor_chain).to include(:exclude_models)
|
27
|
-
end
|
28
|
-
|
29
|
-
context 'with configured model name solr field values' do
|
30
|
-
before do
|
31
|
-
config = IiifPrint::Configuration.new.tap { |c| c.excluded_model_name_solr_field_values = ['Excluded Model', 'Another Excluded Model'] }
|
32
|
-
subject.exclude_models(solr_parameters, config: config)
|
33
|
-
end
|
34
|
-
|
35
|
-
it 'adds the facet fields to solr_parameters with default key' do
|
36
|
-
expect(solr_parameters[:fq]).to be_truthy
|
37
|
-
expect(solr_parameters[:fq]).to(
|
38
|
-
include("-human_readable_type_sim:\"Excluded Model\"", "-human_readable_type_sim:\"Another Excluded Model\"")
|
39
|
-
)
|
40
|
-
end
|
41
|
-
|
42
|
-
context 'with configured model name solr field key' do
|
43
|
-
before do
|
44
|
-
config = IiifPrint::Configuration.new.tap do |c|
|
45
|
-
c.excluded_model_name_solr_field_values = ['ExcludedModel', 'AnotherExcludedModel']
|
46
|
-
c.excluded_model_name_solr_field_key = 'has_model_ssim'
|
47
|
-
end
|
48
|
-
subject.exclude_models(solr_parameters, config: config)
|
49
|
-
end
|
50
|
-
|
51
|
-
it 'adds the facet fields to solr_parameters with configured key' do
|
52
|
-
expect(solr_parameters[:fq]).to be_truthy
|
53
|
-
expect(solr_parameters[:fq]).to(
|
54
|
-
include("-has_model_ssim:\"ExcludedModel\"", "-has_model_ssim:\"AnotherExcludedModel\"")
|
55
|
-
)
|
56
|
-
end
|
57
|
-
end
|
58
|
-
end
|
59
|
-
end
|
60
|
-
end
|
@@ -1,193 +0,0 @@
|
|
1
|
-
require 'spec_helper'
|
2
|
-
|
3
|
-
RSpec.describe IiifPrint::Configuration do
|
4
|
-
let(:config) { described_class.new }
|
5
|
-
|
6
|
-
describe '#ancestory_identifier_function' do
|
7
|
-
subject(:function) { config.ancestory_identifier_function }
|
8
|
-
it "is expected to be a lambda with an arity of one" do
|
9
|
-
expect(function.arity).to eq(1)
|
10
|
-
end
|
11
|
-
|
12
|
-
it "is configurable" do
|
13
|
-
expect do
|
14
|
-
config.ancestory_identifier_function = ->(w) { w.object_id }
|
15
|
-
end.to change { config.ancestory_identifier_function.object_id }
|
16
|
-
end
|
17
|
-
end
|
18
|
-
|
19
|
-
describe '#unique_child_title_generator_function' do
|
20
|
-
subject(:function) { config.unique_child_title_generator_function }
|
21
|
-
|
22
|
-
it "is expected to be a lambda with keyword args" do
|
23
|
-
expect(function.parameters).to eq([[:keyreq, :original_pdf_path],
|
24
|
-
[:keyreq, :image_path],
|
25
|
-
[:keyreq, :parent_work],
|
26
|
-
[:keyreq, :page_number],
|
27
|
-
[:keyreq, :page_padding]])
|
28
|
-
end
|
29
|
-
|
30
|
-
it 'works as originally designed' do
|
31
|
-
work = double(title: ["My Title"], id: '1234')
|
32
|
-
expect(function.call(
|
33
|
-
original_pdf_path: "/hello/world/nice.pdf",
|
34
|
-
image_path: __FILE__,
|
35
|
-
parent_work: work,
|
36
|
-
page_number: 23,
|
37
|
-
page_padding: 5
|
38
|
-
)).to eq("1234 - nice.pdf Page 00024")
|
39
|
-
end
|
40
|
-
|
41
|
-
it "is configurable" do
|
42
|
-
expect do
|
43
|
-
config.unique_child_title_generator_function = ->(**kwargs) { kwargs }
|
44
|
-
end.to change { config.unique_child_title_generator_function.object_id }
|
45
|
-
end
|
46
|
-
end
|
47
|
-
|
48
|
-
describe "#metadata_fields" do
|
49
|
-
subject { config.metadata_fields }
|
50
|
-
|
51
|
-
it { is_expected.to be_a Hash }
|
52
|
-
it "allows for an override" do
|
53
|
-
original = config.metadata_fields
|
54
|
-
config.metadata_fields = { title: {} }
|
55
|
-
expect(config.metadata_fields).not_to eq original
|
56
|
-
end
|
57
|
-
end
|
58
|
-
|
59
|
-
describe "#handle_after_create_fileset" do
|
60
|
-
let(:file_set) { double(FileSet) }
|
61
|
-
let(:user) { double(User) }
|
62
|
-
subject(:called_function) { config.handle_after_create_fileset(file_set, user) }
|
63
|
-
|
64
|
-
context "without configuration" do
|
65
|
-
it "calls IiifPrint::Data.handle_after_create_fileset" do
|
66
|
-
expect(IiifPrint::Data).to receive(:handle_after_create_fileset).with(file_set, user)
|
67
|
-
|
68
|
-
called_function
|
69
|
-
end
|
70
|
-
end
|
71
|
-
|
72
|
-
context "with configuration" do
|
73
|
-
let(:config_func) { ->(_file_set, _user) { :yup } }
|
74
|
-
|
75
|
-
it "calls the given configured lambda" do
|
76
|
-
config.after_create_fileset_handler = config_func
|
77
|
-
expect(IiifPrint::Data).not_to receive(:handle_after_create_fileset)
|
78
|
-
expect(config_func).to receive(:call).with(file_set, user)
|
79
|
-
called_function
|
80
|
-
end
|
81
|
-
end
|
82
|
-
end
|
83
|
-
|
84
|
-
describe '#additional_tesseract_options' do
|
85
|
-
context "by default" do
|
86
|
-
subject { config.additional_tesseract_options }
|
87
|
-
it { is_expected.not_to be_present }
|
88
|
-
end
|
89
|
-
|
90
|
-
it "can be configured" do
|
91
|
-
expect do
|
92
|
-
config.additional_tesseract_options = "-l esperanto"
|
93
|
-
end.to change(config, :additional_tesseract_options)
|
94
|
-
.from("")
|
95
|
-
.to("-l esperanto")
|
96
|
-
end
|
97
|
-
end
|
98
|
-
|
99
|
-
describe '#default_iiif_manifest_version' do
|
100
|
-
subject { config.default_iiif_manifest_version }
|
101
|
-
|
102
|
-
context 'default' do
|
103
|
-
it { is_expected.to eq 2 }
|
104
|
-
end
|
105
|
-
|
106
|
-
context 'when set to empty' do
|
107
|
-
before { config.default_iiif_manifest_version = '' }
|
108
|
-
it { is_expected.to eq 2 }
|
109
|
-
end
|
110
|
-
|
111
|
-
it 'can be set' do
|
112
|
-
expect { config.default_iiif_manifest_version = 3 }
|
113
|
-
.to change(config, :default_iiif_manifest_version)
|
114
|
-
.from(2)
|
115
|
-
.to(3)
|
116
|
-
end
|
117
|
-
end
|
118
|
-
|
119
|
-
describe '#child_work_attributes_function' do
|
120
|
-
subject(:function) { config.child_work_attributes_function }
|
121
|
-
|
122
|
-
it "is expected to be a lambda with keyword args" do
|
123
|
-
expect(function.parameters).to eq([[:keyreq, :parent_work],
|
124
|
-
[:keyreq, :admin_set_id]])
|
125
|
-
end
|
126
|
-
end
|
127
|
-
|
128
|
-
describe "#sort_iiif_manifest_canvases_by" do
|
129
|
-
subject { config.sort_iiif_manifest_canvases_by }
|
130
|
-
|
131
|
-
it { is_expected.to be_a NilClass }
|
132
|
-
it "allows for an override" do
|
133
|
-
original = config.sort_iiif_manifest_canvases_by
|
134
|
-
config.sort_iiif_manifest_canvases_by = :title
|
135
|
-
expect(config.metadata_fields).not_to eq original
|
136
|
-
end
|
137
|
-
end
|
138
|
-
|
139
|
-
describe "#ocr_coords_from_json_function" do
|
140
|
-
subject(:function) { config.ocr_coords_from_json_function }
|
141
|
-
|
142
|
-
it "is expected to be a lambda with one keyword arg and optional args" do
|
143
|
-
expect(function.parameters).to eq([[:keyreq, :file_set_id], [:keyrest]])
|
144
|
-
end
|
145
|
-
end
|
146
|
-
|
147
|
-
describe "#all_text_generator_function" do
|
148
|
-
subject(:function) { config.all_text_generator_function }
|
149
|
-
|
150
|
-
it "is expected to be a lambda with one keyword arg" do
|
151
|
-
expect(function.parameters).to eq([[:keyreq, :object]])
|
152
|
-
end
|
153
|
-
end
|
154
|
-
|
155
|
-
describe "#iiif_metadata_field_presentation_order" do
|
156
|
-
subject { config.iiif_metadata_field_presentation_order }
|
157
|
-
|
158
|
-
it { is_expected.to be_a NilClass }
|
159
|
-
it "allows for an override" do
|
160
|
-
original = config.iiif_metadata_field_presentation_order
|
161
|
-
config.iiif_metadata_field_presentation_order = :title
|
162
|
-
expect(config.iiif_metadata_field_presentation_order).not_to eq original
|
163
|
-
end
|
164
|
-
end
|
165
|
-
|
166
|
-
describe "#questioning_authority_fields" do
|
167
|
-
subject { config.questioning_authority_fields }
|
168
|
-
|
169
|
-
it { is_expected.to be_a Array }
|
170
|
-
context "by default" do
|
171
|
-
it { is_expected.to eq ['rights_statement', 'license'] }
|
172
|
-
end
|
173
|
-
|
174
|
-
it "allows for an override" do
|
175
|
-
expect do
|
176
|
-
config.questioning_authority_fields = ['rights_statement', 'license', 'subject']
|
177
|
-
end.to change(config, :questioning_authority_fields).from(['rights_statement', 'license']).to(['rights_statement', 'license', 'subject'])
|
178
|
-
end
|
179
|
-
end
|
180
|
-
|
181
|
-
describe '#skip_splitting_pdf_files_that_end_with_these_texts' do
|
182
|
-
subject { config.skip_splitting_pdf_files_that_end_with_these_texts }
|
183
|
-
context 'by default' do
|
184
|
-
it { is_expected.to be_empty }
|
185
|
-
end
|
186
|
-
|
187
|
-
context 'is configurable' do
|
188
|
-
before { config.skip_splitting_pdf_files_that_end_with_these_texts = ['.READER.pdf'] }
|
189
|
-
|
190
|
-
it { is_expected.not_to be_empty }
|
191
|
-
end
|
192
|
-
end
|
193
|
-
end
|
@@ -1,245 +0,0 @@
|
|
1
|
-
# encoding: UTF-8
|
2
|
-
|
3
|
-
require 'spec_helper'
|
4
|
-
require 'misc_shared'
|
5
|
-
|
6
|
-
RSpec.describe IiifPrint::Data::WorkDerivatives do
|
7
|
-
include_context "shared setup"
|
8
|
-
|
9
|
-
let(:bare_work) do
|
10
|
-
work = NewspaperPage.new
|
11
|
-
work.title = ['Another one']
|
12
|
-
work.save!
|
13
|
-
work
|
14
|
-
end
|
15
|
-
|
16
|
-
let(:work) do
|
17
|
-
# sample work comes from shared setup, but we need derivative, save...
|
18
|
-
mk_txt_derivative(sample_work)
|
19
|
-
sample_work.save!(validate: false)
|
20
|
-
sample_work
|
21
|
-
end
|
22
|
-
|
23
|
-
let(:adapter) { described_class.new(work) }
|
24
|
-
|
25
|
-
let(:txt1) do
|
26
|
-
registered_dirs = IiifPrint.config.registered_ingest_dirs
|
27
|
-
registered_dirs.push('/tmp') unless registered_dirs.include?('/tmp')
|
28
|
-
file = Tempfile.new(['txt1', '.txt'])
|
29
|
-
file.write('hello')
|
30
|
-
file.flush
|
31
|
-
file
|
32
|
-
end
|
33
|
-
|
34
|
-
let(:txt2) do
|
35
|
-
file = Tempfile.new('txt2.txt')
|
36
|
-
file.write('bye')
|
37
|
-
file.flush
|
38
|
-
file
|
39
|
-
end
|
40
|
-
|
41
|
-
let(:encoded_text) do
|
42
|
-
file = Tempfile.new('txt_encoded.txt', encoding: 'UTF-8')
|
43
|
-
file.write('Gorgonzola Dolce® — on sale for £12.50/kg')
|
44
|
-
file.flush
|
45
|
-
file
|
46
|
-
end
|
47
|
-
|
48
|
-
describe "enumerates available derivatives like hash" do
|
49
|
-
xit "includes expected derivative path for work" do
|
50
|
-
expect(adapter.keys).to include 'txt'
|
51
|
-
end
|
52
|
-
|
53
|
-
xit "can be introspected for quantity of derivatives" do
|
54
|
-
# `size` method without argument is count of derivatives,
|
55
|
-
# functions equivalently to adapter.keys.size
|
56
|
-
expect(adapter.size).to eq adapter.keys.size
|
57
|
-
end
|
58
|
-
|
59
|
-
xit "enumerates expected derivative extension for work" do
|
60
|
-
ext_found = adapter.keys
|
61
|
-
expect(ext_found).to include 'txt'
|
62
|
-
end
|
63
|
-
|
64
|
-
xit "enumerates expected derivative extension for file set" do
|
65
|
-
file_set = work.members.detect { |m| m.is_a? FileSet }
|
66
|
-
adapter = described_class.new(file_set)
|
67
|
-
ext_found = adapter.keys
|
68
|
-
expect(ext_found).to include 'txt'
|
69
|
-
end
|
70
|
-
|
71
|
-
xit "enumerates expected derivative extension for file set id" do
|
72
|
-
file_set = work.members.detect { |m| m.is_a? FileSet }
|
73
|
-
adapter = described_class.new(file_set.id)
|
74
|
-
ext_found = adapter.keys
|
75
|
-
expect(ext_found).to include 'txt'
|
76
|
-
end
|
77
|
-
end
|
78
|
-
|
79
|
-
describe "loads derivatives for a work" do
|
80
|
-
xit "Loads text derivative path" do
|
81
|
-
expect(File.exist?(adapter.path('txt'))).to be true
|
82
|
-
expect(adapter.exist?('txt')).to be true
|
83
|
-
end
|
84
|
-
|
85
|
-
xit "Loads text derivative data" do
|
86
|
-
expect(adapter.data('txt')).to include 'mythical'
|
87
|
-
end
|
88
|
-
|
89
|
-
xit "Handles character encoding on read" do
|
90
|
-
# replace fixture text derivative for work with encoded text
|
91
|
-
adapter.attach(encoded_text.path, 'txt')
|
92
|
-
data = adapter.data('txt')
|
93
|
-
expect(data).to include '—' # em-dash
|
94
|
-
expect(data).to include '£' # gb-pound sign
|
95
|
-
expect(data.encoding.to_s).to eq 'UTF-8'
|
96
|
-
end
|
97
|
-
|
98
|
-
xit "Loads thumbnail derivative data" do
|
99
|
-
mk_thumbnail_derivative(work)
|
100
|
-
# get size by loading data
|
101
|
-
expect(adapter.data('thumbnail').bytes.size).to eq 16_743
|
102
|
-
# get size by File.size via .size method
|
103
|
-
expect(adapter.size('thumbnail')).to eq 16_743
|
104
|
-
end
|
105
|
-
|
106
|
-
xit "Can access jp2 derivative" do
|
107
|
-
mk_jp2_derivative(work)
|
108
|
-
expect(File.exist?(adapter.path('jp2'))).to be true
|
109
|
-
expect(adapter.exist?('jp2')).to be true
|
110
|
-
end
|
111
|
-
end
|
112
|
-
|
113
|
-
describe "create, update, delete derivatives" do
|
114
|
-
xit "will queue derivative file assignment" do
|
115
|
-
adapter = described_class.new(bare_work)
|
116
|
-
adapter.assign(example_gray_jp2)
|
117
|
-
expect(adapter.assigned).to include example_gray_jp2
|
118
|
-
end
|
119
|
-
|
120
|
-
xit "will fail to assign file in non-registered dirs" do
|
121
|
-
adapter = described_class.new(bare_work)
|
122
|
-
# need a non-whitlisted file that exists:
|
123
|
-
bad_path = File.expand_path("../../spec_helper.rb", fixture_path)
|
124
|
-
expect { adapter.assign(bad_path) }.to raise_error(SecurityError)
|
125
|
-
end
|
126
|
-
|
127
|
-
xit "will remove file assignment from queue" do
|
128
|
-
adapter = described_class.new(bare_work)
|
129
|
-
expect(adapter.state).to eq 'empty'
|
130
|
-
adapter.assign(example_gray_jp2)
|
131
|
-
expect(adapter.assigned).to include example_gray_jp2
|
132
|
-
expect(adapter.state).to eq 'dirty'
|
133
|
-
adapter.unassign(example_gray_jp2)
|
134
|
-
expect(adapter.assigned).not_to include example_gray_jp2
|
135
|
-
expect(adapter.state).to eq 'empty'
|
136
|
-
end
|
137
|
-
|
138
|
-
xit "will queue a deletion" do
|
139
|
-
# Given a work with a derivative (txt) already assigned
|
140
|
-
expect(adapter.state).to eq 'saved'
|
141
|
-
# unassigning path...
|
142
|
-
adapter.unassign('txt')
|
143
|
-
# will lead to queued unassignment (intent to delete)...
|
144
|
-
expect(adapter.unassigned).to include 'txt'
|
145
|
-
# and a 'dirty' adapter state (unflushed changes):
|
146
|
-
expect(adapter.state).to eq 'dirty'
|
147
|
-
end
|
148
|
-
|
149
|
-
xit "will flush a removal and addition on commit!" do
|
150
|
-
# Given a work with a derivative (txt) already assigned
|
151
|
-
expect(adapter.keys).to include 'txt'
|
152
|
-
expect(adapter.keys).not_to include 'jp2'
|
153
|
-
# unassigning path...
|
154
|
-
adapter.unassign('txt')
|
155
|
-
# and assigning another attachment:
|
156
|
-
adapter.assign(example_gray_jp2)
|
157
|
-
# ...committing these will flush the changes (synchronously):
|
158
|
-
adapter.commit!
|
159
|
-
expect(adapter.keys).not_to include 'txt'
|
160
|
-
expect(adapter.keys).to include 'jp2'
|
161
|
-
expect(adapter.size('jp2')).to eq 27_703
|
162
|
-
end
|
163
|
-
|
164
|
-
xit "can attach derivative from file" do
|
165
|
-
expect(adapter.keys).not_to include 'jp2'
|
166
|
-
adapter.attach(example_gray_jp2, 'jp2')
|
167
|
-
expect(adapter.exist?('jp2')).to be true
|
168
|
-
expect(adapter.path('jp2')).not_to be nil
|
169
|
-
expect(File.size(adapter.path('jp2'))).to eq File.size(example_gray_jp2)
|
170
|
-
expect(adapter.keys).to include 'jp2'
|
171
|
-
d_path = path_factory.derivative_path_for_reference(adapter.fileset_id, 'jp2')
|
172
|
-
expect(adapter.values).to include d_path
|
173
|
-
end
|
174
|
-
|
175
|
-
xit "can replace aderivative with new attachment" do
|
176
|
-
adapter.attach(txt1.path, 'txt')
|
177
|
-
expect(adapter.data('txt')).to eq 'hello'
|
178
|
-
adapter.attach(txt2.path, 'txt')
|
179
|
-
expect(adapter.data('txt')).to eq 'bye'
|
180
|
-
end
|
181
|
-
|
182
|
-
xit "can delete an attached derivative" do
|
183
|
-
adapter.attach(txt1.path, 'txt')
|
184
|
-
expect(adapter.keys).to include 'txt'
|
185
|
-
expect(adapter.data('txt')).to eq 'hello'
|
186
|
-
adapter.delete('txt')
|
187
|
-
expect(adapter.path('txt')).to be nil
|
188
|
-
expect(adapter.keys).not_to include 'txt'
|
189
|
-
end
|
190
|
-
|
191
|
-
xit "persists log of attachment to RDBMS" do
|
192
|
-
adapter.assign(txt1.path)
|
193
|
-
result = IiifPrint::DerivativeAttachment.find_by(
|
194
|
-
fileset_id: adapter.fileset.id,
|
195
|
-
path: txt1.path,
|
196
|
-
destination_name: 'txt'
|
197
|
-
)
|
198
|
-
expect(result).not_to be_nil
|
199
|
-
end
|
200
|
-
|
201
|
-
xit "persists a log of path relation to primary file" do
|
202
|
-
# this is an integration test by practical necessity, with
|
203
|
-
# WorkFiles adapting a bare work with no fileset.
|
204
|
-
work_files = IiifPrint::Data::WorkFiles.of(bare_work)
|
205
|
-
work_files.assign(example_gray_jp2)
|
206
|
-
adapter = work_files.derivatives
|
207
|
-
adapter.assign(txt1.path)
|
208
|
-
result = IiifPrint::IngestFileRelation.find_by(
|
209
|
-
derivative_path: txt1.path,
|
210
|
-
file_path: example_gray_jp2
|
211
|
-
)
|
212
|
-
expect(result).not_to be_nil
|
213
|
-
end
|
214
|
-
|
215
|
-
xit "commits queued derivatives" do
|
216
|
-
IiifPrint::IngestFileRelation.where(file_path: example_gray_jp2).delete_all
|
217
|
-
work_files = IiifPrint::Data::WorkFiles.of(bare_work)
|
218
|
-
work_files.assign(example_gray_jp2)
|
219
|
-
adapter = work_files.derivatives
|
220
|
-
adapter.assign(txt1.path)
|
221
|
-
expect(File.exist?(txt1.path)).to be true
|
222
|
-
expect(adapter.keys.size).to eq 0
|
223
|
-
# we need a fileset, saved with import_url, attached to work:
|
224
|
-
fileset = valid_file_set
|
225
|
-
fileset.import_url = 'file://' + example_gray_jp2
|
226
|
-
fileset.save!
|
227
|
-
bare_work.members.push(fileset)
|
228
|
-
bare_work.save!
|
229
|
-
fileset.reload
|
230
|
-
expect(fileset.member_of[0].id).to eq bare_work.id
|
231
|
-
# with a new adapter instance...
|
232
|
-
adapter2 = described_class.of(bare_work)
|
233
|
-
# call .commit_queued! with our fileset...
|
234
|
-
expect(File.exist?(txt1.path)).to be true
|
235
|
-
adapter2.commit_queued!(fileset)
|
236
|
-
# ...which should result in saved, reloaded derivative...
|
237
|
-
expect(adapter2.keys.size).to eq 1
|
238
|
-
expect(File.size(adapter2.values[0])).to eq File.size(txt1.path)
|
239
|
-
# ...also found via Hyrax::DerviativePath:
|
240
|
-
found = Hyrax::DerivativePath.derivatives_for_reference(fileset.id)
|
241
|
-
expect(found.size).to eq 1
|
242
|
-
expect(File.size(found[0])).to eq File.size(txt1.path)
|
243
|
-
end
|
244
|
-
end
|
245
|
-
end
|