Skip to content

Commit

Permalink
Override IiifPrint
Browse files Browse the repository at this point in the history 
Temporary patches until the changes are pushed into IiifPrint.

See #728
and child_work_attributes_function should call Array.wrap not to_a #729
  • Loading branch information
@laritakr
laritakr committed Aug 5, 2024
1 parent 5e92ad1 commit c88e0b7
Show file tree
Hide file tree
Showing 2 changed files with 141 additions and 91 deletions.
198 changes: 107 additions & 91 deletions app/jobs/iiif_print/child_works_from_pdf_job_decorator.rb
View file
Original file line number Diff line number Diff line change
Expand Up @@ -3,8 +3,9 @@
# OVERRIDE: This is a wholecloth override of
# https://github.com/scientist-softserv/iiif_print/pull/295/files We're doing this because the IIIF
# Print build is now taking 2 hours; which means a 4 hour loop to bring updates from IIIF Print.
require 'iiif_print/jobs/application_job'

module IiifPrint
# rubocop:disable Metrics/LineLength
module ChildWorksFromPdfJobDecorator
##
# Break a pdf into individual pages
Expand All @@ -13,26 +14,29 @@ module ChildWorksFromPdfJobDecorator
# @param pdf_paths: [<Array => String>] paths to pdfs
# @param user: [User]
# @param admin_set_id: [<String>]
def perform(candidate_for_parency, pdf_paths, user, admin_set_id, *)
# rubocop:disable Metrics/MethodLength
def perform(id, pdf_paths, user, admin_set_id, *)
candidate_for_parency = IiifPrint.find_by(id: id)

##
# We know that we have cases where parent_work is nil, this will definitely raise an
# exception; which is fine because we were going to do it later anyway.
@parent_work = if candidate_for_parency.work?
pdf_file_set = nil
candidate_for_parency
else
# We likely have a file set
pdf_file_set = candidate_for_parency
IiifPrint.parent_for(candidate_for_parency)
end
pdf_file_set = nil
candidate_for_parency
else
# We likely have a file set
pdf_file_set = candidate_for_parency
IiifPrint.parent_for(candidate_for_parency)
end
@child_admin_set_id = admin_set_id
child_model = @parent_work.iiif_print_config.pdf_split_child_model

# When working with remote files, we have put the PDF file into the correct path before submitting this job.
# However, there seem to be cases where we still don't have the file when we get here, so to be sure, we
# re-do the same command that was previously used to prepare the file path. If the file is already here, it
# simply returns the path, but if not it will copy the file there, giving us one more chance to have what we need.
pdf_paths = [Hyrax::WorkingDirectory.find_or_retrieve(pdf_file_set.files.first.id, pdf_file_set.id, pdf_paths.first)] if pdf_file_set
pdf_paths = [Hyrax::WorkingDirectory.find_or_retrieve(pdf_file_set.original_file.id, pdf_file_set.id, pdf_paths.first)] if pdf_file_set
# handle each input pdf (when input is a file set, we will only have one).
pdf_paths.each do |original_pdf_path|
split_pdf(original_pdf_path, user, child_model, pdf_file_set)
Expand All @@ -45,103 +49,115 @@ def perform(candidate_for_parency, pdf_paths, user, admin_set_id, *)
# @param child_model: [<String>] child model
IiifPrint::Jobs::CreateRelationshipsJob.set(wait: 10.minutes).perform_later(
user: user,
parent_id: @parent_work.id,
parent_id: @parent_work.id.to_s,
parent_model: @parent_work.class.to_s,
child_model: child_model.to_s
)

# TODO: clean up image_files and pdf_paths
end
# rubocop:enable Metrics/MethodLength

private

def split_pdf(original_pdf_path, user, child_model, pdf_file_set)
image_files = @parent_work.iiif_print_config.pdf_splitter_service.call(original_pdf_path, file_set: pdf_file_set)

# give as much info as possible if we don't have image files to work with.
if image_files.blank?
raise "#{@parent_work.class} (ID=#{@parent_work.id} " \
"to_param:#{@parent_work.to_param}) " \
"original_pdf_path #{original_pdf_path.inspect} " \
"pdf_file_set #{pdf_file_set.inspect}"
end
# rubocop:disable Metrics/ParameterLists
# rubocop:disable Metrics/MethodLength
def split_pdf(original_pdf_path, user, child_model, pdf_file_set)
user = User.find_by_user_key(user) unless user.is_a?(User)
image_files = @parent_work.iiif_print_config.pdf_splitter_service.call(original_pdf_path, file_set: pdf_file_set)

# give as much info as possible if we don't have image files to work with.
if image_files.blank?
raise "#{@parent_work.class} (ID=#{@parent_work.id} " \
"to_param:#{@parent_work.to_param}) " \
"original_pdf_path #{original_pdf_path.inspect} " \
"pdf_file_set #{pdf_file_set.inspect}"
end

@split_from_pdf_id = pdf_file_set.nil? ? nil : pdf_file_set.id
prepare_import_data(original_pdf_path, image_files, user)

# submit the job to create all the child works for one PDF
# @param [User] user
# @param [Hash<String => String>] titles
# @param [Hash<String => String>] resource_types (optional)
# @param [Array<String>] uploaded_files Hyrax::UploadedFile IDs
# @param [Hash] attributes attributes to apply to all works, including :model
# @param [Hyrax::BatchCreateOperation] operation
operation = Hyrax::BatchCreateOperation.create!(
user: user,
operation_type: "PDF Batch Create"
@split_from_pdf_id = pdf_file_set&.id.to_s
prepare_import_data(original_pdf_path, image_files, user)

# submit the job to create all the child works for one PDF
# @param [User] user
# @param [Hash<String => String>] titles
# @param [Hash<String => String>] resource_types (optional)
# @param [Array<String>] uploaded_files Hyrax::UploadedFile IDs
# @param [Hash] attributes attributes to apply to all works, including :model
# @param [Hyrax::BatchCreateOperation] operation
operation = Hyrax::BatchCreateOperation.create!(
user: user,
operation_type: "PDF Batch Create"
)
BatchCreateJob.perform_later(user,
@child_work_titles,
@resource_types,
@uploaded_files,
attributes.merge!(model: child_model.to_s, split_from_pdf_id: @split_from_pdf_id).with_indifferent_access,
operation)
end
# rubocop:enable Metrics/MethodLength
# rubocop:enable Metrics/ParameterLists

# rubocop:disable Metrics/MethodLength
def prepare_import_data(original_pdf_path, image_files, user)
@uploaded_files = []
@child_work_titles = {}
@resource_types = {}
number_of_pages_in_pdf = image_files.size
image_files.each_with_index do |image_path, page_number|
file_id = create_uploaded_file(user, image_path).to_s

child_title = IiifPrint.config.unique_child_title_generator_function.call(
original_pdf_path: original_pdf_path,
image_path: image_path,
parent_work: @parent_work,
page_number: page_number,
page_padding: number_of_digits(nbr: number_of_pages_in_pdf)
)
BatchCreateJob.perform_later(user,
@child_work_titles,
{},
@uploaded_files,
attributes.merge!(model: child_model.to_s, split_from_pdf_id: @split_from_pdf_id).with_indifferent_access,
operation)
end
# rubocop:enable Metrics/ParameterLists

def prepare_import_data(original_pdf_path, image_files, user)
@uploaded_files = []
@child_work_titles = {}
number_of_pages_in_pdf = image_files.size
image_files.each_with_index do |image_path, page_number|
file_id = create_uploaded_file(user, image_path).to_s

child_title = IiifPrint.config.unique_child_title_generator_function.call(
original_pdf_path: original_pdf_path,
image_path: image_path,
parent_work: @parent_work,
page_number: page_number,
page_padding: number_of_digits(nbr: number_of_pages_in_pdf)
)

@uploaded_files << file_id
@child_work_titles[file_id] = child_title
# save child work info to create the member relationships
PendingRelationship.create!(child_title: child_title,
parent_id: @parent_work.id,
child_order: child_title,
parent_model: @parent_work.class,
child_model: @parent_work.iiif_print_config.pdf_split_child_model,
file_id: @split_from_pdf_id)

begin
# Clean up the temporary image path.
File.rm_f(image_path) if File.exist?(image_path)
rescue StandardError => e
Rails.logger.debug("Unable to delete #{image_path}; with error #{e}.")
# If we can't delete, let's move on. Maybe it was already cleaned-up.
end

@resource_types[file_id] = resource_types
@uploaded_files << file_id
@child_work_titles[file_id] = child_title
# save child work info to create the member relationships
PendingRelationship.create!(child_title: child_title,
parent_id: @parent_work.id,
child_order: child_title,
parent_model: @parent_work.class,
child_model: @parent_work.iiif_print_config.pdf_split_child_model,
file_id: @split_from_pdf_id)

begin
# Clean up the temporary image path.
FileUtils.rm_f(image_path) if File.exist?(image_path)
rescue
# If we can't delete, let's move on. Maybe it was already cleaned-up.
end
end
# rubocop:enable Metrics/MethodLength
end
# rubocop:enable Metrics/MethodLength

def number_of_digits(nbr:)
nbr.to_s.size
end
def number_of_digits(nbr:)
nbr.to_s.size
end

def create_uploaded_file(user, path)
# TODO: Could we create a remote path?
uf = Hyrax::UploadedFile.new
uf.user_id = user.id
uf.file = CarrierWave::SanitizedFile.new(path)
uf.save!
uf.id
end
def create_uploaded_file(user, path)
# TODO: Could we create a remote path?
uf = Hyrax::UploadedFile.new
uf.user_id = user.try(:id) || user
uf.file = CarrierWave::SanitizedFile.new(path)
uf.save!
uf.id
end

# TODO: what attributes do we need to fill in from the parent work? What about AllinsonFlex?
def attributes
IiifPrint.config.child_work_attributes_function.call(parent_work: @parent_work, admin_set_id: @child_admin_set_id)
end
def attributes
IiifPrint.config.child_work_attributes_function.call(parent_work: @parent_work,
admin_set_id: @child_admin_set_id)
end

# TODO: Does this method need to be configurable?
def resource_types
@parent_work.try(:resource_type)
end
end
# rubocop:enable Metrics/LineLength
end
Expand Down
34 changes: 34 additions & 0 deletions config/initializers/iiif_print.rb
View file
Original file line number Diff line number Diff line change
Expand Up @@ -95,6 +95,40 @@
collection: {}
}

config.child_work_attributes_function = lambda do |parent_work:, admin_set_id:|
embargo = parent_work.embargo
lease = parent_work.lease
embargo_params = {}
lease_params = {}
visibility_params = {}

if embargo
embargo_params = {
visibility: 'embargo',
visibility_after_embargo: embargo.visibility_after_embargo,
visibility_during_embargo: embargo.visibility_during_embargo,
embargo_release_date: embargo.embargo_release_date
}
elsif lease
lease_params = {
visibility: 'lease',
visibility_after_lease: lease.visibility_after_lease,
visibility_during_lease: lease.visibility_during_lease,
lease_release_date: lease.lease_release_date
}
else
visibility_params = { visibility: parent_work.visibility.to_s }
end

params = {
admin_set_id: admin_set_id.to_s,
creator: Array.wrap(parent_work.creator),
rights_statement: Array.wrap(parent_work.rights_statement),
is_child: true
}

params.merge!(embargo_params).merge!(lease_params).merge!(visibility_params)
end
# rubocop:enable Metrics/LineLength
end

Expand Down

0 comments on commit c88e0b7

Please sign in to comment.