Job for importing an object from Purl (Stanford)

Still need to wrap this with webmock or similar and invoke it via a
script

app/jobs/import_work_from_purl_job.rb

@@ -0,0 +1,77 @@
+require 'stanford'
+# Import works from Purl/stacks services at Stanford
+#
+# Example usage:
+#       log = Hyrax::Operation.create!(user: current_user,
+#                                      operation_type: "Import Purl Metadata")
+#       ImportWorkFromPurlJob.perform_later(current_user,
+#                                           'abcd1234xxxx',
+#                                           log)
+class ImportWorkFromPurlJob < ActiveJob::Base
+  queue_as :ingest
+
+  before_enqueue do |job|
+    log = job.arguments.last
+    log.pending_job(self)
+  end
+
+  # This copies metadata from the passed in attribute to all of the works that
+  # are members of the given upload set
+  # @param [User] user
+  # @param [String] druid
+  # @param [Hyrax::Operation] log
+  def perform(user, druid, log)
+    xml = Stanford::Importer::PurlRetriever.get(druid)
+    parser = Stanford::Importer::PurlParser.new(xml)
+    attributes = process_attributes(parser.attributes)
+    model = model_to_create(attributes)
+
+    CreateWorkJob.perform_now(user, model, attributes, log)
+  end
+
+  private
+
+    def process_attributes(attributes)
+      # We're pruning off :form_of_work, :record_origin, :created_attributes, :identifiers
+      attributes = attributes.slice(*attributes_to_keep)
+      # rename :location to :based_near
+      attributes[:based_near] = attributes.delete(:location)
+
+      process_collection(attributes)
+      filenames = attributes.delete(:files)
+      attributes[:remote_files] = filenames.map do |name|
+        { url: "https://stacks.stanford.edu/file/druid:#{attributes[:id]}/#{name}",
+          file_name: name }
+      end
+
+      attributes
+    end
+
+    class_attribute :attributes_to_keep
+    self.attributes_to_keep = [:title,
+                               :description,
+                               :subject,
+                               :language,
+                               :resource_type,
+                               :location,
+                               :rights,
+                               :visibility,
+                               :id,
+                               :collection,
+                               :files]
+
+    def process_collection(attributes)
+      # rename :collection to :member_of_collection_ids
+      collection = attributes.delete(:collection)
+
+      Collection.create!(collection) unless Collection.exists?(collection[:id])
+      attributes[:member_of_collection_ids] = [collection[:id]]
+    end
+
+    # Override this method if you have a different rubric for choosing the model
+    # @param [Hash] attributes
+    # @return String the model to create
+    def model_to_create(attributes)
+      Hyrax.config.model_to_create.call(attributes)
+    end
+end

app/middleware/account_elevator.rb

@@ -10,7 +10,12 @@ def parse_tenant_name(request)
 
   def self.switch!(cname)
     account = Account.find_by(cname: Account.canonical_cname(cname))
-
-    Apartment::Tenant.switch!(account.tenant)
+    if account
+      Apartment::Tenant.switch!(account.tenant)
+    elsif Account.any?
+      raise "No tenant found for #{cname}"
+    else
+      logger.info "It looks like we're in single tenant mode. No tenant found for #{cname}"
+    end
   end
 end

bin/import_from_purl

@@ -14,34 +14,37 @@ def validate_druids!(druids)
   exit(1)
 end
 
-def validate_imagepath!(imagepath)
-  return if imagepath
+def validate_username!(username)
+  user = User.find_by_user_key(username)
+  return user if user
   usage
-  $stderr.puts 'Image directory was left blank. No images will be ingested'
+  $stderr.puts 'username was left blank.'
   exit(1)
 end
 
-def load_rails
+def load_rails(hostname)
   puts 'Loading environment...'
   require File.expand_path('../../config/environment', __FILE__)
   require 'stanford'
-  puts 'Starting import...'
+  AccountElevator.switch!(hostname)
 end
 
-def main(hostname, imagepath, druids)
+def main(hostname, username, druids)
   validate_hostname!(hostname)
   validate_druids!(druids)
-  validate_imagepath!(imagepath)
-  load_rails
+  load_rails(hostname)
+  user = validate_username!(username)
 
-  AccountElevator.switch!(hostname)
-  size = Stanford::Importer::PurlImporter.new(imagepath, druids).import_all
+  druids.each do |druid|
+    log = Hyrax::Operation.create!(user: user, operation_type: "Import Purl Metadata")
+    ImportWorkFromPurlJob.perform_later(user, druid, log)
+  end
 
-  puts "Imported #{size} records."
+  puts "Enqueued #{druids.size} import jobs."
 end
 
 def usage
-  $stderr.puts "Usage: #{$PROGRAM_NAME} <hostname> <image directory> <druids to import>"
+  $stderr.puts "Usage: #{$PROGRAM_NAME} <hostname> <username> <druids to import>"
 end
 
 main(ARGV[0], ARGV[1], ARGV.drop(2))

config/environments/development.rb

@@ -26,6 +26,8 @@
     config.cache_store = :null_store
   end
 
+  config.active_job.queue_adapter = :sidekiq
+
   # Don't care if the mailer can't send.
   config.action_mailer.raise_delivery_errors = false
 

lib/stanford/importer.rb

@@ -4,5 +4,6 @@ module Importer
     autoload :ModsParser
     autoload :PurlImporter
     autoload :PurlParser
+    autoload :PurlRetriever
   end
 end

lib/stanford/importer/purl_importer.rb

@@ -43,11 +43,7 @@ def files(attributes)
       private
 
         def retrieve(druid)
-          conn.get("/#{druid}.xml").body
-        end
-
-        def conn
-          @conn ||= Faraday.new(url: 'https://purl.stanford.edu')
+          PurlRetriever.get(druid)
         end
     end
   end

lib/stanford/importer/purl_retriever.rb

@@ -0,0 +1,21 @@
+module Stanford
+  module Importer
+    class PurlRetriever
+      def self.get(druid)
+        new(druid).get
+      end
+
+      def initialize(druid)
+        @druid = druid
+      end
+
+      def get
+        conn.get("/#{@druid}.xml").body
+      end
+
+      def conn
+        @conn ||= Faraday.new(url: 'https://purl.stanford.edu')
+      end
+    end
+  end
+end

spec/jobs/import_work_from_purl_job_spec.rb

@@ -0,0 +1,20 @@
+require 'rails_helper'
+
+RSpec.describe ImportWorkFromPurlJob do
+  let(:user) { create(:user) }
+  let(:log) { Hyrax::Operation.create!(user: user, operation_type: "Import Purl Metadata") }
+  let(:druid) { 'bc390xk2647' }
+  before do
+    if ActiveFedora::Base.exists? druid
+      ActiveFedora::Base.find(druid).destroy(eradicate: true)
+    end
+    Hyrax::Workflow::WorkflowImporter.load_workflows
+    Hyrax::PermissionTemplate.create!(admin_set_id: Hyrax::DefaultAdminSetActor::DEFAULT_ID, workflow_name: 'default')
+  end
+  it "works" do
+    expect do
+      described_class.perform_now(user, druid, log)
+    end.to change { GenericWork.count }.by(1)
+      .and change { FileSet.count }.by(2)
+  end
+end

spec/spec_helper.rb

@@ -95,6 +95,9 @@
   #     --seed 1234
   config.order = :random
 
+  # Give us a full backtrace on Travis-CI
+  config.backtrace_exclusion_patterns = [] if ENV['CI']
+
   # Seed global randomization in this process using the `--seed` CLI option.
   # Setting this allows you to use `--seed` to deterministically reproduce
   # test failures related to randomization by passing the same `--seed` value