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
@@ -1,11 +1,13 @@
|
|
1
1
|
class CreateIiifPrintPendingRelationships < ActiveRecord::Migration[5.1]
|
2
2
|
def change
|
3
|
-
|
4
|
-
|
5
|
-
|
6
|
-
|
7
|
-
|
3
|
+
unless table_exists?(:iiif_print_pending_relationships)
|
4
|
+
create_table :iiif_print_pending_relationships do |t|
|
5
|
+
t.string :child_title, null: false
|
6
|
+
t.string :parent_id, null: false
|
7
|
+
t.string :child_order, null: false
|
8
|
+
t.timestamps
|
9
|
+
end
|
10
|
+
add_index :iiif_print_pending_relationships, :parent_id
|
8
11
|
end
|
9
|
-
add_index :iiif_print_pending_relationships, :parent_id
|
10
12
|
end
|
11
13
|
end
|
@@ -1,7 +1,7 @@
|
|
1
1
|
class AddModelDetailsToIiifPrintPendingRelationships < ActiveRecord::Migration[5.2]
|
2
2
|
def change
|
3
|
-
add_column :iiif_print_pending_relationships, :parent_model, :string
|
4
|
-
add_column :iiif_print_pending_relationships, :child_model, :string
|
5
|
-
add_column :iiif_print_pending_relationships, :file_id, :string
|
3
|
+
add_column :iiif_print_pending_relationships, :parent_model, :string unless column_exists?(:iiif_print_pending_relationships, :parent_model)
|
4
|
+
add_column :iiif_print_pending_relationships, :child_model, :string unless column_exists?(:iiif_print_pending_relationships, :child_model)
|
5
|
+
add_column :iiif_print_pending_relationships, :file_id, :string unless column_exists?(:iiif_print_pending_relationships, :file_id)
|
6
6
|
end
|
7
7
|
end
|
data/iiif_print.gemspec
CHANGED
@@ -18,7 +18,7 @@ Gem::Specification.new do |spec|
|
|
18
18
|
IiifPrint is a gem (Rails "engine") for Hyrax-based digital repository applications to support displaying parent/child works in the same viewer (Universal Viewer) and the ability to search OCR from the parent work to the child work(s). IiifPring was originally based off of the samvera-labs Newspaper gem.
|
19
19
|
SUMMARY
|
20
20
|
spec.license = 'Apache-2.0'
|
21
|
-
spec.files = `git ls-files`.split($OUTPUT_RECORD_SEPARATOR)
|
21
|
+
spec.files = `git ls-files`.split($OUTPUT_RECORD_SEPARATOR).select { |f| File.dirname(f) !~ %r{\A"?spec\/?} && f != 'bin/rails' }
|
22
22
|
spec.test_files = spec.files.grep(%r{^(test|spec|features)/})
|
23
23
|
spec.add_dependency 'blacklight_iiif_search', '>= 1.0', '< 3.0'
|
24
24
|
spec.add_dependency 'derivative-rodeo', "~> 0.5"
|
@@ -7,7 +7,11 @@ module IiifPrint
|
|
7
7
|
class_attribute :target_extension, default: nil
|
8
8
|
|
9
9
|
def initialize(file_set)
|
10
|
-
@file_set = file_set
|
10
|
+
@file_set = if file_set.is_a?(Hyrax::FileMetadata)
|
11
|
+
Hyrax.query_service.find_by(id: file_set.file_set_id)
|
12
|
+
else
|
13
|
+
file_set
|
14
|
+
end
|
11
15
|
@dest_path = nil
|
12
16
|
@source_path = nil
|
13
17
|
@source_meta = nil
|
@@ -26,7 +30,10 @@ module IiifPrint
|
|
26
30
|
# @return [Boolean]
|
27
31
|
def valid?
|
28
32
|
# @note We are taking a shortcut because currently we are only concerned about images.
|
29
|
-
|
33
|
+
# @TODO: verify if this works for ActiveFedora and if so, remove commented code.
|
34
|
+
# If not, modify to use adapter.
|
35
|
+
# file_set.class.image_mime_types.include?(file_set.mime_type)
|
36
|
+
file_set.original_file.image?
|
30
37
|
end
|
31
38
|
|
32
39
|
def derivative_path_factory
|
@@ -110,5 +117,9 @@ module IiifPrint
|
|
110
117
|
# intermediate -> PDF
|
111
118
|
im_convert
|
112
119
|
end
|
120
|
+
|
121
|
+
def mime_type_for(extension)
|
122
|
+
Marcel::MimeType.for extension: extension
|
123
|
+
end
|
113
124
|
end
|
114
125
|
end
|
@@ -103,10 +103,10 @@ module IiifPrint
|
|
103
103
|
def file_set_id
|
104
104
|
return document['id'] if document.file_set?
|
105
105
|
|
106
|
-
file_set_ids = document['
|
106
|
+
file_set_ids = document['member_ids_ssim']
|
107
107
|
raise "#{self.class}: NO FILE SET ID" if file_set_ids.blank?
|
108
108
|
|
109
|
-
# Since a parent work's `
|
109
|
+
# Since a parent work's `member_ids_ssim` can contain child work ids as well as file set ids,
|
110
110
|
# this will ensure that the file set id is indeed a `FileSet`
|
111
111
|
file_set_ids.detect { |id| SolrDocument.find(id).file_set? }
|
112
112
|
end
|
@@ -25,9 +25,9 @@ module IiifPrint
|
|
25
25
|
# rubocop:enable Naming/PredicateName
|
26
26
|
def show_parents_only(solr_parameters)
|
27
27
|
query = if blacklight_params["include_child_works"] == 'true'
|
28
|
-
|
28
|
+
IiifPrint.solr_construct_query(is_child_bsi: 'true')
|
29
29
|
else
|
30
|
-
|
30
|
+
IiifPrint.solr_construct_query(is_child_bsi: nil)
|
31
31
|
end
|
32
32
|
solr_parameters[:fq] += [query]
|
33
33
|
end
|
@@ -3,6 +3,36 @@ module IiifPrint
|
|
3
3
|
class Configuration
|
4
4
|
attr_writer :after_create_fileset_handler
|
5
5
|
|
6
|
+
attr_writer :ingest_queue_name
|
7
|
+
##
|
8
|
+
# @return [Symbol, Proc]
|
9
|
+
def ingest_queue_name
|
10
|
+
return @ingest_queue_name if @ingest_queue_name.present?
|
11
|
+
if defined?(Hyrax)
|
12
|
+
Hyrax.config.ingest_queue_name
|
13
|
+
elsif defined?(Bulkrax) && Bulkrax.config.respond_to?(:ingest_queue_name)
|
14
|
+
Bulkrax.config.ingest_queue_name
|
15
|
+
else
|
16
|
+
:ingest
|
17
|
+
end
|
18
|
+
end
|
19
|
+
|
20
|
+
attr_writer :persistence_adapter
|
21
|
+
def persistence_adapter
|
22
|
+
@persistence_adapter || default_persistence_adapter
|
23
|
+
end
|
24
|
+
|
25
|
+
def default_persistence_adapter
|
26
|
+
# There's probably some configuration of Hyrax we could use to better refine this; but it's
|
27
|
+
# likely a reasonable guess. The main goal is to not break existing implementations and
|
28
|
+
# maintain an upgrade path.
|
29
|
+
if Gem::Version.new(Hyrax::VERSION) >= Gem::Version.new('6.0.0')
|
30
|
+
IiifPrint::PersistenceLayer::ValkyrieAdapter
|
31
|
+
else
|
32
|
+
IiifPrint::PersistenceLayer::ActiveFedoraAdapter
|
33
|
+
end
|
34
|
+
end
|
35
|
+
|
6
36
|
# @param file_set [FileSet]
|
7
37
|
# @param user [User]
|
8
38
|
def handle_after_create_fileset(file_set, user)
|
@@ -14,11 +44,14 @@ module IiifPrint
|
|
14
44
|
end
|
15
45
|
|
16
46
|
attr_writer :ancestory_identifier_function
|
17
|
-
# The function, with arity 1, that receives a work and returns it's identifier
|
18
|
-
# of object ancestry.
|
47
|
+
# The function, with arity 1, that receives a work and returns it's identifier (as a string) for
|
48
|
+
# the purposes of object ancestry.
|
49
|
+
#
|
19
50
|
# @return [Proc]
|
20
51
|
def ancestory_identifier_function
|
21
|
-
|
52
|
+
# If the work.id is nil, keep it nil. Otherwise cast that id to a string; to deal with the
|
53
|
+
# `Valkyrie::ID`.
|
54
|
+
@ancestory_identifier_function ||= ->(work) { work.id&.to_s }
|
22
55
|
end
|
23
56
|
|
24
57
|
attr_writer :excluded_model_name_solr_field_values
|
@@ -158,17 +191,44 @@ module IiifPrint
|
|
158
191
|
attr_writer :child_work_attributes_function
|
159
192
|
##
|
160
193
|
# Here we allow for customization of the child work attributes
|
194
|
+
# rubocop:disable Metrics/MethodLength, Metrics/BlockLength
|
161
195
|
def child_work_attributes_function
|
162
196
|
@child_work_attributes_function ||= lambda do |parent_work:, admin_set_id:|
|
163
|
-
|
197
|
+
embargo = parent_work.embargo
|
198
|
+
lease = parent_work.lease
|
199
|
+
embargo_params = {}
|
200
|
+
lease_params = {}
|
201
|
+
visibility_params = {}
|
202
|
+
|
203
|
+
if embargo
|
204
|
+
embargo_params = {
|
205
|
+
visibility: 'embargo',
|
206
|
+
visibility_after_embargo: embargo.visibility_after_embargo,
|
207
|
+
visibility_during_embargo: embargo.visibility_during_embargo,
|
208
|
+
embargo_release_date: embargo.embargo_release_date
|
209
|
+
}
|
210
|
+
elsif lease
|
211
|
+
lease_params = {
|
212
|
+
visibility: 'lease',
|
213
|
+
visibility_after_lease: lease.visibility_after_lease,
|
214
|
+
visibility_during_lease: lease.visibility_during_lease,
|
215
|
+
lease_release_date: lease.lease_release_date
|
216
|
+
}
|
217
|
+
else
|
218
|
+
visibility_params = { visibility: parent_work.visibility.to_s }
|
219
|
+
end
|
220
|
+
|
221
|
+
params = {
|
164
222
|
admin_set_id: admin_set_id.to_s,
|
165
223
|
creator: parent_work.creator.to_a,
|
166
224
|
rights_statement: parent_work.rights_statement.to_a,
|
167
|
-
visibility: parent_work.visibility.to_s,
|
168
225
|
is_child: true
|
169
226
|
}
|
227
|
+
|
228
|
+
params.merge!(embargo_params).merge!(lease_params).merge!(visibility_params)
|
170
229
|
end
|
171
230
|
end
|
231
|
+
# rubocop:enable Metrics/MethodLength, Metrics/BlockLength
|
172
232
|
|
173
233
|
attr_writer :sort_iiif_manifest_canvases_by
|
174
234
|
##
|
@@ -15,9 +15,9 @@ module IiifPrint
|
|
15
15
|
# get the fileset from that id
|
16
16
|
return FileSet.find(@work) if @work.is_a?(String)
|
17
17
|
# if "work" context is a FileSet, not actual work, return it
|
18
|
-
return @work if @work.is_a? FileSet
|
18
|
+
return @work if @work.is_a?(Hyrax::FileSet) || @work.is_a?(FileSet)
|
19
19
|
# in most cases, get from work's members:
|
20
|
-
filesets = @work.members.select { |m| m.is_a? FileSet }
|
20
|
+
filesets = @work.members.select { |m| m.is_a?(Hyrax::FileSet) || m.is_a?(FileSet) }
|
21
21
|
filesets.empty? ? nil : filesets[0]
|
22
22
|
end
|
23
23
|
end
|
@@ -239,7 +239,7 @@ module IiifPrint
|
|
239
239
|
# of the first assigned file path for single-file work.
|
240
240
|
work_file = parent
|
241
241
|
return if work_file.nil?
|
242
|
-
work_files = work_file
|
242
|
+
work_files = IiifPrint.parent_for(work_file)
|
243
243
|
return if work_files.nil?
|
244
244
|
work_files.assigned[0]
|
245
245
|
else
|
data/lib/iiif_print/engine.rb
CHANGED
@@ -11,6 +11,15 @@ module IiifPrint
|
|
11
11
|
class Engine < ::Rails::Engine
|
12
12
|
isolate_namespace IiifPrint
|
13
13
|
|
14
|
+
config.eager_load_paths += %W[#{config.root}/app/transactions]
|
15
|
+
|
16
|
+
initializer 'requires' do
|
17
|
+
require 'hyrax/transactions/iiif_print_container_decorator'
|
18
|
+
require 'iiif_print/persistence_layer'
|
19
|
+
require 'iiif_print/persistence_layer/active_fedora_adapter' if defined?(ActiveFedora)
|
20
|
+
require 'iiif_print/persistence_layer/valkyrie_adapter' if defined?(Valkyrie)
|
21
|
+
end
|
22
|
+
|
14
23
|
# rubocop:disable Metrics/BlockLength
|
15
24
|
config.to_prepare do
|
16
25
|
require "iiif_print/jobs/create_relationships_job"
|
@@ -37,16 +46,51 @@ module IiifPrint
|
|
37
46
|
IiifPrint::PluggableDerivativeService
|
38
47
|
)
|
39
48
|
|
49
|
+
Hyrax.publisher.subscribe(IiifPrint::Listener.new) if Hyrax.respond_to?(:publisher)
|
50
|
+
|
40
51
|
Hyrax::IiifManifestPresenter.prepend(IiifPrint::IiifManifestPresenterBehavior)
|
41
52
|
Hyrax::IiifManifestPresenter::Factory.prepend(IiifPrint::IiifManifestPresenterFactoryBehavior)
|
42
53
|
Hyrax::ManifestBuilderService.prepend(IiifPrint::ManifestBuilderServiceBehavior)
|
43
54
|
Hyrax::Renderers::FacetedAttributeRenderer.prepend(Hyrax::Renderers::FacetedAttributeRendererDecorator)
|
44
55
|
Hyrax::WorksControllerBehavior.prepend(IiifPrint::WorksControllerBehaviorDecorator)
|
56
|
+
"Hyrax::Transactions::Steps::DeleteAllFileSets".safe_constantize&.prepend(Hyrax::Transactions::Steps::DeleteAllFileSetsDecorator)
|
57
|
+
# Hyku::WorksControllerBehavior was introduced in Hyku v6.0.0+. Yes we don't depend on Hyku,
|
58
|
+
# but this allows us to do minimal Hyku antics with IiifPrint.
|
59
|
+
'Hyku::WorksControllerBehavior'.safe_constantize&.prepend(IiifPrint::WorksControllerBehaviorDecorator)
|
60
|
+
|
61
|
+
Hyrax::FileSetPresenter.prepend(IiifPrint::FileSetPresenterDecorator)
|
45
62
|
Hyrax::WorkShowPresenter.prepend(IiifPrint::WorkShowPresenterDecorator)
|
46
63
|
Hyrax::IiifHelper.prepend(IiifPrint::IiifHelperDecorator)
|
47
64
|
|
48
|
-
|
49
|
-
|
65
|
+
if ActiveModel::Type::Boolean.new.cast(ENV.fetch('HYRAX_VALKYRIE', false))
|
66
|
+
# Newer versions of Hyrax favor `Hyrax::Indexers::FileSetIndexer` and deprecate
|
67
|
+
# `Hyrax::ValkyrieFileSetIndexer`.
|
68
|
+
'Hyrax::Indexers::FileSetIndexer'.safe_constantize&.prepend(IiifPrint::FileSetIndexer)
|
69
|
+
|
70
|
+
# Versions 3.0+ of Hyrax have `Hyrax::ValkyrieFileSetIndexer` so we want to decorate that as
|
71
|
+
# well. We want to use the elsif construct because later on Hyrax::ValkyrieFileSetIndexer
|
72
|
+
# inherits from Hyrax::Indexers::FileSetIndexer and only implements:
|
73
|
+
# `def initialize(*args); super; end`
|
74
|
+
'Hyrax::ValkyrieFileSetIndexer'.safe_constantize&.prepend(IiifPrint::FileSetIndexer)
|
75
|
+
|
76
|
+
# Newer versions of Hyrax favor `Hyrax::Indexers::PcdmObjectIndexer` and deprecate
|
77
|
+
# `Hyrax::ValkyrieWorkIndexer`
|
78
|
+
indexers = Hyrax.config.curation_concerns.map do |concern|
|
79
|
+
"#{concern}ResourceIndexer".safe_constantize
|
80
|
+
end
|
81
|
+
indexers.each { |indexer| indexer.prepend(IiifPrint::ChildWorkIndexer) }
|
82
|
+
|
83
|
+
# Versions 3.0+ of Hyrax have `Hyrax::ValkyrieWorkIndexer` so we want to decorate that as
|
84
|
+
# well. We want to use the elsif construct because later on Hyrax::ValkyrieWorkIndexer
|
85
|
+
# inherits from Hyrax::Indexers::PcdmObjectIndexer and only implements:
|
86
|
+
# `def initialize(*args); super; end`
|
87
|
+
'Hyrax::ValkyrieWorkIndexer'.safe_constantize&.prepend(IiifPrint::ChildWorkIndexer)
|
88
|
+
else
|
89
|
+
# The ActiveFedora::Base indexer for FileSets
|
90
|
+
Hyrax::FileSetIndexer.prepend(IiifPrint::FileSetIndexer)
|
91
|
+
# The ActiveFedora::Base indexer for Works
|
92
|
+
Hyrax::WorkIndexer.prepend(IiifPrint::ChildWorkIndexer)
|
93
|
+
end
|
50
94
|
|
51
95
|
::BlacklightIiifSearch::IiifSearchResponse.prepend(IiifPrint::IiifSearchResponseDecorator)
|
52
96
|
::BlacklightIiifSearch::IiifSearchAnnotation.prepend(IiifPrint::BlacklightIiifSearch::AnnotationDecorator)
|
@@ -7,9 +7,9 @@ module IiifPrint
|
|
7
7
|
|
8
8
|
def show_parents_only(solr_parameters)
|
9
9
|
query = if blacklight_params["include_child_works"] == 'true'
|
10
|
-
|
10
|
+
IiifPrint.solr_construct_query(is_child_bsi: 'true')
|
11
11
|
else
|
12
|
-
|
12
|
+
IiifPrint.solr_construct_query(is_child_bsi: nil)
|
13
13
|
end
|
14
14
|
solr_parameters[:fq] += [query]
|
15
15
|
end
|
@@ -53,7 +53,10 @@ module IiifPrint
|
|
53
53
|
render_cmd = opj_command
|
54
54
|
|
55
55
|
# Run the generated command to make derivative file at @dest_path
|
56
|
-
`#{render_cmd}`
|
56
|
+
data = `#{render_cmd}`
|
57
|
+
|
58
|
+
# Create Hyrax::FileMetadata object for the derivatives (if Valkyrie)
|
59
|
+
IiifPrint.copy_derivatives_from_data_store(stream: data, directives: { url: file_set.id.to_s, container: 'service_file', mime_type: mime_type_for(target_extension) })
|
57
60
|
|
58
61
|
# Clean up any intermediate files or symlinks used during creation
|
59
62
|
cleanup_intermediate
|
@@ -3,6 +3,7 @@ module IiifPrint
|
|
3
3
|
#
|
4
4
|
# - {.ancestor_ids_for}
|
5
5
|
# - {.descendent_member_ids_for}
|
6
|
+
# - {.ancestor_identifier_for}
|
6
7
|
#
|
7
8
|
# The ancestor and descendent_file_sets are useful for ensuring we index together related items.
|
8
9
|
# For example, when I have a work that is a book, and one file set per page of that book, when I
|
@@ -15,13 +16,21 @@ module IiifPrint
|
|
15
16
|
#
|
16
17
|
# @param object [#in_works] An object that responds to #in_works
|
17
18
|
# @return [Array<String>]
|
19
|
+
#
|
20
|
+
# @note For those implementing their own lineage service, verify that you are not returning
|
21
|
+
# an array of
|
18
22
|
def self.ancestor_ids_for(object)
|
19
23
|
ancestor_ids ||= []
|
20
|
-
|
24
|
+
# Yes, we're fetching the works, then compressing those into identifiers. Because in the case
|
25
|
+
# of slugs, we need not the identifier, but the slug as the id.
|
26
|
+
IiifPrint.object_in_works(object).each do |work|
|
21
27
|
ancestor_ids << ancestry_identifier_for(work)
|
22
|
-
ancestor_ids += ancestor_ids_for(work) if work.is_child
|
28
|
+
ancestor_ids += ancestor_ids_for(work) if work.respond_to?(:is_child) && work.is_child
|
23
29
|
end
|
24
|
-
|
30
|
+
# We must convert these to strings as Valkyrie's identifiers will be cast to hashes when we
|
31
|
+
# attempt to write the SolrDocument. Also, per documentation we return an Array of strings, not
|
32
|
+
# an Array that might include Valkyrie::ID objects.
|
33
|
+
ancestor_ids.flatten.compact.uniq.map(&:to_s)
|
25
34
|
end
|
26
35
|
|
27
36
|
##
|
@@ -43,6 +52,8 @@ module IiifPrint
|
|
43
52
|
# https://github.com/samvera/hyrax/blob/2b807fe101176d594129ef8a8fe466d3d03a372b/app/indexers/hyrax/work_indexer.rb#L15-L18
|
44
53
|
# for "clarification" of the comingling of file_set_ids and member_ids
|
45
54
|
def self.descendent_member_ids_for(object)
|
55
|
+
return unless object.respond_to?(:member_ids)
|
56
|
+
|
46
57
|
# enables us to return parents when searching for child OCR
|
47
58
|
#
|
48
59
|
# https://github.com/samvera/hydra-works/blob/c9b9dd0cf11de671920ba0a7161db68ccf9b7f6d/lib/hydra/works/models/concerns/work_behavior.rb#L90-L92
|
@@ -50,10 +61,12 @@ module IiifPrint
|
|
50
61
|
# The Hydara::Works implementation of file_set_ids is "members.select(&:file_set?).map(&:id)";
|
51
62
|
# so no sense doing `object.file_set_ids + object.member_ids`
|
52
63
|
file_set_ids = object.member_ids
|
53
|
-
object
|
54
|
-
file_set_ids += descendent_member_ids_for(child)
|
64
|
+
IiifPrint.object_ordered_works(object)&.each do |child|
|
65
|
+
file_set_ids += Array.wrap(descendent_member_ids_for(child))
|
55
66
|
end
|
56
|
-
|
67
|
+
# We must convert these to strings as Valkyrie's identifiers will be cast to hashes when we
|
68
|
+
# attempt to write the SolrDocument.
|
69
|
+
file_set_ids.flatten.uniq.compact.map(&:to_s)
|
57
70
|
end
|
58
71
|
class << self
|
59
72
|
alias descendent_file_set_ids_for descendent_member_ids_for
|
@@ -24,7 +24,9 @@ module IiifPrint
|
|
24
24
|
# JP2 source, and whether we have color or grayscale material.
|
25
25
|
def convert_cmd
|
26
26
|
template = use_color? ? COLOR_PDF_CMD : GRAY_PDF_CMD
|
27
|
-
format(template, source_file: @source_path, out_file: @dest_path)
|
27
|
+
data = format(template, source_file: @source_path, out_file: @dest_path)
|
28
|
+
IiifPrint.copy_derivatives_from_data_store(stream: data, directives: { url: file_set.id.to_s, container: 'service_file', mime_type: mime_type_for(target_extension) })
|
29
|
+
data
|
28
30
|
end
|
29
31
|
|
30
32
|
def create_derivatives(filename)
|
@@ -0,0 +1,189 @@
|
|
1
|
+
module IiifPrint
|
2
|
+
module PersistenceLayer
|
3
|
+
class ActiveFedoraAdapter < AbstractAdapter
|
4
|
+
##
|
5
|
+
# @param object [ActiveFedora::Base]
|
6
|
+
# @return [Array<SolrDocument>]
|
7
|
+
def self.object_in_works(object)
|
8
|
+
object.in_works
|
9
|
+
end
|
10
|
+
|
11
|
+
##
|
12
|
+
# @param object [ActiveFedora::Base]
|
13
|
+
# @return [Array<SolrDocument>]
|
14
|
+
def self.object_ordered_works(object)
|
15
|
+
object.ordered_works
|
16
|
+
end
|
17
|
+
|
18
|
+
##
|
19
|
+
# @param work_type [Class<ActiveFedora::Base>]
|
20
|
+
# @return indexer for the given :work_type
|
21
|
+
def self.decorate_with_adapter_logic(work_type:)
|
22
|
+
work_type.send(:include, IiifPrint::SetChildFlag) unless work_type.included_modules.include?(IiifPrint::SetChildFlag)
|
23
|
+
work_type.indexer
|
24
|
+
end
|
25
|
+
|
26
|
+
##
|
27
|
+
# @param work_type [Class<ActiveFedora::Base>]
|
28
|
+
# @return indexer for the given :work_type
|
29
|
+
def self.decorate_form_with_adapter_logic(work_type:)
|
30
|
+
work_type.indexer
|
31
|
+
end
|
32
|
+
|
33
|
+
##
|
34
|
+
# Return the immediate parent of the given :file_set.
|
35
|
+
#
|
36
|
+
# @param file_set [FileSet]
|
37
|
+
# @return [#work?, Hydra::PCDM::Work]
|
38
|
+
# @return [NilClass] when no parent is found.
|
39
|
+
def self.parent_for(file_set)
|
40
|
+
# fallback to Fedora-stored relationships if work's aggregation of
|
41
|
+
# file set is not indexed in Solr
|
42
|
+
file_set.parent || file_set.member_of.find(&:work?)
|
43
|
+
end
|
44
|
+
|
45
|
+
##
|
46
|
+
# Return the parent's parent of the given :file_set.
|
47
|
+
#
|
48
|
+
# @param file_set [FileSet]
|
49
|
+
# @return [#work?, Hydra::PCDM::Work]
|
50
|
+
# @return [NilClass] when no grand parent is found.
|
51
|
+
def self.grandparent_for(file_set)
|
52
|
+
parent_of_file_set = parent_for(file_set)
|
53
|
+
# HACK: This is an assumption about the file_set structure, namely that an image page split from
|
54
|
+
# a PDF is part of a file set that is a child of a work that is a child of a single work. That
|
55
|
+
# is, it only has one grand parent. Which is a reasonable assumption for IIIF Print but is not
|
56
|
+
# valid when extended beyond IIIF Print. That is GenericWork does not have a parent method but
|
57
|
+
# does have a parents method.
|
58
|
+
parent_of_file_set.try(:parent_works).try(:first) ||
|
59
|
+
parent_of_file_set.try(:parents).try(:first) ||
|
60
|
+
parent_of_file_set&.member_of&.find(&:work?)
|
61
|
+
end
|
62
|
+
|
63
|
+
def self.solr_construct_query(*args)
|
64
|
+
if defined?(Hyrax::SolrQueryBuilderService)
|
65
|
+
Hyrax::SolrQueryBuilderService.construct_query(*args)
|
66
|
+
else
|
67
|
+
ActiveFedora::SolrQueryBuilder.construct_query(*args)
|
68
|
+
end
|
69
|
+
end
|
70
|
+
|
71
|
+
def self.clean_for_tests!
|
72
|
+
super do
|
73
|
+
ActiveFedora::Cleaner.clean!
|
74
|
+
end
|
75
|
+
end
|
76
|
+
|
77
|
+
def self.solr_query(query, **args)
|
78
|
+
if defined?(ActiveFedora::SolrService)
|
79
|
+
ActiveFedora::SolrService.query(query, **args)
|
80
|
+
else
|
81
|
+
Hyrax::SolrService.query(query, **args)
|
82
|
+
end
|
83
|
+
end
|
84
|
+
|
85
|
+
def self.solr_name(field_name)
|
86
|
+
if defined?(Hyrax) && Hyrax.config.respond_to?(:index_field_mapper)
|
87
|
+
Hyrax.config.index_field_mapper.solr_name(field_name.to_s)
|
88
|
+
else
|
89
|
+
::ActiveFedora.index_field_mapper.solr_name(field_name.to_s)
|
90
|
+
end
|
91
|
+
end
|
92
|
+
|
93
|
+
##
|
94
|
+
# @param file_set [Object]
|
95
|
+
# @param work [Object]
|
96
|
+
# @param model [Class] The class name for which we'll split children.
|
97
|
+
def self.destroy_children_split_from(file_set:, work:, model:, **_args)
|
98
|
+
# look first for children by the file set id they were split from
|
99
|
+
children = model.where(split_from_pdf_id: file_set.id)
|
100
|
+
if children.blank?
|
101
|
+
# find works where file name and work `to_param` are both in the title
|
102
|
+
children = model.where(title: file_set.label).where(title: work.to_param)
|
103
|
+
end
|
104
|
+
return if children.blank?
|
105
|
+
children.each do |rcd|
|
106
|
+
rcd.destroy(eradicate: true)
|
107
|
+
end
|
108
|
+
true
|
109
|
+
end
|
110
|
+
|
111
|
+
def self.pdf?(file_set)
|
112
|
+
file_set.class.pdf_mime_types.include?(file_set.mime_type)
|
113
|
+
end
|
114
|
+
|
115
|
+
##
|
116
|
+
# Add a child record as a member of a parent record
|
117
|
+
#
|
118
|
+
# @param model [child_record] an ActiveFedora::Base model
|
119
|
+
# @param model [parent_record] an ActiveFedora::Base model
|
120
|
+
# @return [TrueClass]
|
121
|
+
def self.create_relationship_between(child_record:, parent_record:)
|
122
|
+
return true if parent_record.ordered_members.to_a.include?(child_record)
|
123
|
+
parent_record.ordered_members << child_record
|
124
|
+
true
|
125
|
+
end
|
126
|
+
|
127
|
+
##
|
128
|
+
# find a work by title
|
129
|
+
# We should only find one, but there is no guarantee of that and `:where` returns an array.
|
130
|
+
#
|
131
|
+
# @param title [String]
|
132
|
+
# @param model [String] an ActiveFedora::Base model
|
133
|
+
def self.find_by_title_for(title:, model:)
|
134
|
+
work_type = model.constantize
|
135
|
+
|
136
|
+
work_type.where(title: title)
|
137
|
+
end
|
138
|
+
|
139
|
+
##
|
140
|
+
# find a work or file_set
|
141
|
+
#
|
142
|
+
# @param id [String]
|
143
|
+
# @return [Array<ActiveFedora::Base]
|
144
|
+
def self.find_by(id:)
|
145
|
+
ActiveFedora::Base.find(id)
|
146
|
+
end
|
147
|
+
|
148
|
+
##
|
149
|
+
# save a work
|
150
|
+
#
|
151
|
+
# @param object [Array<ActiveFedora::Base]
|
152
|
+
def self.save(object:)
|
153
|
+
object.save!
|
154
|
+
end
|
155
|
+
|
156
|
+
##
|
157
|
+
# reindex an array of works and their file_sets
|
158
|
+
#
|
159
|
+
# @param objects [Array<ActiveFedora::Base]
|
160
|
+
# @return [TrueClass]
|
161
|
+
def self.index_works(objects:)
|
162
|
+
objects.each do |work|
|
163
|
+
work.update_index
|
164
|
+
work.file_sets.each(&:update_index) if work.respond_to?(:file_sets)
|
165
|
+
end
|
166
|
+
true
|
167
|
+
end
|
168
|
+
|
169
|
+
##
|
170
|
+
# does nothing for ActiveFedora;
|
171
|
+
# allows valkyrie works to have an extra step to create the Hyrax::Metadata objects.
|
172
|
+
#
|
173
|
+
# @param []
|
174
|
+
# @return [TrueClass]
|
175
|
+
def self.copy_derivatives_from_data_store(*)
|
176
|
+
true
|
177
|
+
end
|
178
|
+
|
179
|
+
##
|
180
|
+
# Extract text from the derivatives
|
181
|
+
#
|
182
|
+
# @param [FileSet] an ActiveFedora fileset
|
183
|
+
# @return [String] Text from fileset's file
|
184
|
+
def self.extract_text_for(file_set:)
|
185
|
+
IiifPrint.config.all_text_generator_function.call(object: file_set) || ''
|
186
|
+
end
|
187
|
+
end
|
188
|
+
end
|
189
|
+
end
|