hyrax 1.1.1 → 2.0.0.beta1

data/app/jobs/hyrax/grant_read_to_members_job.rb

@@ -0,0 +1,24 @@
+module Hyrax
+  # Grants read access for the supplied user for the members attached to a work
+  class GrantReadToMembersJob < ApplicationJob
+    queue_as Hyrax.config.ingest_queue_name
+
+    # @param [ActiveFedora::Base] work - the work object
+    # @param [String] user_key - the user to add
+    def perform(work, user_key)
+      # Iterate over ids because reifying objects is slow.
+      file_set_ids(work).each do |file_set_id|
+        # Call this synchronously, since we're already in a job
+        GrantReadJob.perform_now(file_set_id, user_key)
+      end
+    end
+
+    private
+
+      # Filter the member ids and return only the FileSet ids (filter out child works)
+      # @return [Array<String>] the file set ids
+      def file_set_ids(work)
+        ::FileSet.search_with_conditions(id: work.member_ids).map(&:id)
+      end
+  end
+end

data/app/jobs/hyrax/revoke_edit_from_members_job.rb

@@ -0,0 +1,23 @@
+module Hyrax
+  # Revokes edit access for the supplied user for the members attached to a work
+  class RevokeEditFromMembersJob < ApplicationJob
+    queue_as Hyrax.config.ingest_queue_name
+
+    # @param [ActiveFedora::Base] work - the work object
+    # @param [String] user_key - the user to remove
+    def perform(work, user_key)
+      # Iterate over ids because reifying objects is slow.
+      file_set_ids(work).each do |file_set_id|
+        RevokeEditJob.perform_now(file_set_id, user_key)
+      end
+    end
+
+    private
+
+      # Filter the member ids and return only the FileSet ids (filter out child works)
+      # @return [Array<String>] the file set ids
+      def file_set_ids(work)
+        ::FileSet.search_with_conditions(id: work.member_ids).map(&:id)
+      end
+  end
+end

data/app/jobs/hyrax/revoke_edit_job.rb

@@ -0,0 +1,14 @@
+module Hyrax
+  # Revokes the user's edit access on the provided FileSet
+  class RevokeEditJob < ApplicationJob
+    queue_as Hyrax.config.ingest_queue_name
+
+    # @param [String] file_set_id - the identifier of the object to revoke access from
+    # @param [String] user_key - the user to remove
+    def perform(file_set_id, user_key)
+      file_set = ::FileSet.find(file_set_id)
+      file_set.edit_users -= [user_key]
+      file_set.save!
+    end
+  end
+end

data/app/jobs/import_export_job.rb

@@ -0,0 +1,31 @@
+require 'open3'
+
+class ImportExportJob < ActiveJob::Base
+  include Open3
+
+  # @param resource [String] uri of the resource
+  # @param mode [String] "import" or "export" mode
+  # @param dir [String] location of export
+  # TODO: Support BagIt profile
+  def perform(resource:, mode: 'export', dir: Hyrax.config.bagit_dir)
+    command = "java -jar #{Hyrax.config.import_export_jar_file_path} --mode #{mode} --resource #{resource} --dir #{dir}"
+    internal_call(command)
+  end
+
+  private
+
+    def internal_call(command)
+      stdin, stdout, stderr, wait_thr = popen3(command)
+      begin
+        out = stdout.read
+        err = stderr.read
+        exit_status = wait_thr.value
+        raise "Fedora command failed: \"#{command}\"\n#{err}" unless exit_status.success?
+        out
+      ensure
+        stdin.close
+        stdout.close
+        stderr.close
+      end
+    end
+end

data/app/jobs/import_url_job.rb

@@ -1,12 +1,12 @@
 require 'uri'
-require 'tempfile'
+require 'tmpdir'
 require 'browse_everything/retriever'
 
 # Given a FileSet that has an import_url property,
 # download that file and put it into Fedora
 # Called by AttachFilesToWorkJob (when files are uploaded to s3)
 # and CreateWithRemoteFilesActor when files are located in some other service.
-class ImportUrlJob < ActiveJob::Base
+class ImportUrlJob < Hyrax::ApplicationJob
   queue_as Hyrax.config.ingest_queue_name
 
   before_enqueue do |job|
@@ -19,39 +19,43 @@ class ImportUrlJob < ActiveJob::Base
   def perform(file_set, operation)
     operation.performing!
     user = User.find_by_user_key(file_set.depositor)
-
-    Tempfile.open(file_set.id.tr('/', '_')) do |f|
-      copy_remote_file(file_set, f)
-
+    uri = URI(file_set.import_url)
+    copy_remote_file(uri) do |f|
       # reload the FileSet once the data is copied since this is a long running task
       file_set.reload
 
-      # We invoke the FileSetActor in a synchronous way so that this tempfile is available
-      # when IngestFileJob is invoked. If it was asynchronous the IngestFileJob may be invoked
-      # on a machine that did not have this temp file on it's file system.
+      # FileSetActor operates synchronously so that this tempfile is available.
+      # If asynchronous, the job might be invoked on a machine that did not have this temp file on its file system!
       # NOTE: The return status may be successful even if the content never attaches.
-      if Hyrax::Actors::FileSetActor.new(file_set, user).create_content(f, 'original_file', false)
-        # send message to user on download success
-        Hyrax.config.callback.run(:after_import_url_success, file_set, user)
+      if Hyrax::Actors::FileSetActor.new(file_set, user).create_content(f)
         operation.success!
       else
+        # send message to user on download failure
         Hyrax.config.callback.run(:after_import_url_failure, file_set, user)
         operation.fail!(file_set.errors.full_messages.join(' '))
       end
     end
   end
 
-  protected
-
-    def copy_remote_file(file_set, f)
-      f.binmode
-      # download file from url
-      uri = URI(file_set.import_url)
-      spec = { 'url' => uri }
-      retriever = BrowseEverything::Retriever.new
-      retriever.retrieve(spec) do |chunk|
-        f.write(chunk)
+  private
+
+    # Download file from uri, yields a block with a file in a temporary directory.
+    # It is important that the file on disk has the same file name as the URL,
+    # because when the file in added into Fedora the file name will get persisted in the
+    # metadata.
+    # @param uri [URI] the uri of the file to download
+    # @yield [IO] the stream to write to
+    def copy_remote_file(uri)
+      filename = File.basename(uri.path)
+      Dir.mktmpdir do |dir|
+        File.open(File.join(dir, filename), 'wb') do |f|
+          retriever = BrowseEverything::Retriever.new
+          retriever.retrieve('url' => uri) do |chunk|
+            f.write(chunk)
+          end
+          f.rewind
+          yield f
+        end
       end
-      f.rewind
     end
 end

data/app/jobs/ingest_job.rb

@@ -0,0 +1,18 @@
+class IngestJob < Hyrax::ApplicationJob
+  queue_as Hyrax.config.ingest_queue_name
+
+  after_perform do |job|
+    # We want the lastmost Hash, if any.
+    opts = job.arguments.reverse.detect { |x| x.is_a? Hash } || {}
+    wrapper = job.arguments.first
+    ContentNewVersionEventJob.perform_later(wrapper.file_set, wrapper.user) if opts[:notification]
+  end
+
+  # @param [JobIoWrapper] wrapper
+  # @param [Boolean] notification send the user a notification, used in after_perform callback
+  # @see 'config/initializers/hyrax_callbacks.rb'
+  # rubocop:disable Lint/UnusedMethodArgument
+  def perform(wrapper, notification: false)
+    wrapper.ingest_file
+  end
+end

data/app/jobs/ingest_local_file_job.rb

@@ -1,4 +1,4 @@
-class IngestLocalFileJob < ActiveJob::Base
+class IngestLocalFileJob < Hyrax::ApplicationJob
   queue_as Hyrax.config.ingest_queue_name
 
   # @param [FileSet] file_set

data/app/jobs/inherit_permissions_job.rb

@@ -1,6 +1,6 @@
 # A job to apply work permissions to all contained files set
 #
-class InheritPermissionsJob < ActiveJob::Base
+class InheritPermissionsJob < Hyrax::ApplicationJob
   # Perform the copy from the work to the contained filesets
   #
   # @param work containing access level and filesets

data/app/jobs/resolrize_job.rb

@@ -1,4 +1,4 @@
-class ResolrizeJob < ActiveJob::Base
+class ResolrizeJob < Hyrax::ApplicationJob
   def perform
     ActiveFedora::Base.reindex_everything
   end

data/app/jobs/visibility_copy_job.rb

@@ -1,18 +1,27 @@
-class VisibilityCopyJob < ActiveJob::Base
+# Responsible for copying the following attributes from the work to each file in the file_sets
+#
+# * visibility
+# * lease
+# * embargo
+class VisibilityCopyJob < Hyrax::ApplicationJob
+  # @api public
+  # @param [#file_sets, #visibility, #lease, #embargo] work - a Work model
   def perform(work)
     work.file_sets.each do |file|
       file.visibility = work.visibility # visibility must come first, because it can clear an embargo/lease
-      if work.lease
-        file.build_lease unless file.lease
-        file.lease.attributes = work.lease.attributes.except('id')
-        file.lease.save
-      end
-      if work.embargo
-        file.build_embargo unless file.embargo
-        file.embargo.attributes = work.embargo.attributes.except('id')
-        file.embargo.save
-      end
+      copy_visibility_modifier(work: work, file: file, modifier: :lease)
+      copy_visibility_modifier(work: work, file: file, modifier: :embargo)
       file.save!
     end
   end
+
+  private
+
+    def copy_visibility_modifier(work:, file:, modifier:)
+      work_modifier = work.public_send(modifier)
+      return unless work_modifier
+      file.public_send("build_#{modifier}") unless file.public_send(modifier)
+      file.public_send(modifier).attributes = work_modifier.attributes.except('id')
+      file.public_send(modifier).save
+    end
 end

data/app/models/admin_set.rb

@@ -1,5 +1,111 @@
-# There is an interplay between an AdminSet and a PermissionTemplate.
-# @see Hyrax::AdminSetBehavior for further discussion
+# There is an interplay between an AdminSet and a PermissionTemplate. Given
+# that AdminSet is an ActiveFedora::Base and PermissionTemplate is ActiveRecord::Base
+# we don't have the usual :has_many or :belongs_to methods to assist in defining that
+# relationship. However, from a conceptual standpoint:
+#
+# * An AdminSet has_one :permission_tempate
+# * A PermissionTemplate belongs_to :admin_set
+#
+# When an object is added as a member of an AdminSet, the AdminSet's associated
+# PermissionTemplate is applied to that object (e.g. some of the object's attributes
+# are updated as per the rules of the permission template)
+#
+# @see Hyrax::PermissionTemplate
+# @see Hyrax::AdminSetService
+# @see Hyrax::Forms::PermissionTemplateForm for validations and creation process
+# @see Hyrax::DefaultAdminSetActor
+# @see Hyrax::ApplyPermissionTemplateActor
 class AdminSet < ActiveFedora::Base
-  include Hyrax::AdminSetBehavior
+  include Hydra::AccessControls::WithAccessRight
+  include Hyrax::Noid
+  include Hyrax::HumanReadableType
+  include Hyrax::HasRepresentative
+
+  DEFAULT_ID = 'admin_set/default'.freeze
+  DEFAULT_TITLE = ['Default Admin Set'].freeze
+  DEFAULT_WORKFLOW_NAME = Hyrax.config.default_active_workflow_name
+
+  validates_with Hyrax::HasOneTitleValidator
+  class_attribute :human_readable_short_description
+  self.indexer = Hyrax::AdminSetIndexer
+  property :title, predicate: ::RDF::Vocab::DC.title do |index|
+    index.as :stored_searchable, :facetable
+  end
+  property :description, predicate: ::RDF::Vocab::DC.description do |index|
+    index.as :stored_searchable
+  end
+
+  property :creator, predicate: ::RDF::Vocab::DC11.creator do |index|
+    index.as :symbol
+  end
+
+  has_many :members,
+           predicate: ::RDF::Vocab::DC.isPartOf,
+           class_name: 'ActiveFedora::Base'
+
+  before_destroy :check_if_not_default_set, :check_if_empty
+  after_destroy :destroy_permission_template
+
+  def self.default_set?(id)
+    id == DEFAULT_ID
+  end
+
+  def default_set?
+    self.class.default_set?(id)
+  end
+
+  # Creates the default AdminSet and an associated PermissionTemplate with workflow
+  def self.find_or_create_default_admin_set_id
+    unless exists?(DEFAULT_ID)
+      Hyrax::AdminSetCreateService.create_default_admin_set(admin_set_id: DEFAULT_ID, title: DEFAULT_TITLE)
+    end
+    DEFAULT_ID
+  end
+
+  def to_s
+    title.present? ? title : 'No Title'
+  end
+
+  # @api public
+  # A bit of an analogue for a `has_one :admin_set` as it crosses from Fedora to the DB
+  # @return [Hyrax::PermissionTemplate]
+  # @raise [ActiveRecord::RecordNotFound]
+  def permission_template
+    Hyrax::PermissionTemplate.find_by!(admin_set_id: id)
+  end
+
+  # @api public
+  #
+  # @return [Sipity::Workflow]
+  # @raise [ActiveRecord::RecordNotFound]
+  def active_workflow
+    Sipity::Workflow.find_active_workflow_for(admin_set_id: id)
+  end
+
+  # Calculate and update who should have edit access based on who
+  # has "manage" access in the PermissionTemplateAccess
+  def update_access_controls!
+    update!(edit_users: permission_template.agent_ids_for(access: 'manage', agent_type: 'user'),
+            edit_groups: permission_template.agent_ids_for(access: 'manage', agent_type: 'group'))
+  end
+
+  private
+
+    def destroy_permission_template
+      permission_template.destroy
+    rescue ActiveRecord::RecordNotFound
+      true
+    end
+
+    def check_if_empty
+      return true if members.empty?
+      errors[:base] << I18n.t('hyrax.admin.admin_sets.delete.error_not_empty')
+      throw :abort
+    end
+
+    def check_if_not_default_set
+      return true unless default_set?
+      errors[:base] << I18n.t('hyrax.admin.admin_sets.delete.error_default_set')
+      throw :abort
+    end
 end

data/app/models/batch_upload_item.rb

@@ -2,8 +2,10 @@
 # It should never actually be persisted in the repository.
 # The properties on this form should be copied to a real work type.
 class BatchUploadItem < ActiveFedora::Base
-  include ::Hyrax::BasicMetadata
   include Hyrax::WorkBehavior
+  # This must come after the WorkBehavior because it finalizes the metadata
+  # schema (by adding accepts_nested_attributes)
+  include ::Hyrax::BasicMetadata
 
   attr_accessor :payload_concern # a Class name: what is this a batch of?
 

data/app/models/checksum_audit_log.rb

@@ -1,19 +1,66 @@
 class ChecksumAuditLog < ActiveRecord::Base
-  def self.get_audit_log(id, path, version_uri)
-    ChecksumAuditLog.find_or_create_by(file_set_id: id, file_id: path, version: version_uri)
+  def failed?
+    !passed?
   end
 
-  # Check to see if there are previous passing logs that we can delete
-  # we want to keep the first passing event after a failure, the most current passing event,
-  # and all failures so that this table doesn't grow too large
-  # Simple way (a little naive): if the last 2 were passing, delete the first one
-  def self.prune_history(id, path)
-    list = logs_for(id, path).limit(2)
-    return if list.size <= 1 || list[0].pass != 1 || list[1].pass != 1
-    list[0].destroy
+  # Only the latest rows for a given file_set_id/checked_uri pair.
+  # Uses a join, so you might have to be careful combining. You would
+  # normally combine this with other conditions, this alone will return
+  # LOTS of records.
+  def self.latest_checks
+    # one crazy SQL trick to get the latest for each fileset/checked_uri combo
+    # where there's no other self-join created_at greater -- only the greatest.
+    joins("LEFT JOIN checksum_audit_logs c2 ON
+            (checksum_audit_logs.file_set_id = c2.file_set_id AND
+             checksum_audit_logs.checked_uri = c2.checked_uri AND
+             checksum_audit_logs.created_at < c2.created_at)")
+      .where("c2.id is NULL")
+      .order("created_at desc, id desc")
   end
 
-  def self.logs_for(id, path)
-    ChecksumAuditLog.where(file_set_id: id, file_id: path).order('created_at desc, id desc')
+  # From all ChecksumAuditLogs related to this file set, returns only
+  # the LATEST for each file_set_id/checked_uri pair.
+  def self.latest_for_file_set_id(file_set_id)
+    latest_checks.where(file_set_id: file_set_id)
+  end
+
+  # Responsible for coordinating the creation (and possible pruning) of checksum log entries
+  #
+  # @param [Boolean] passed - Did the fixity test pass?
+  # @param [#to_s] checked_uri - Which URI did we check?
+  # @param [String] file_set_id - The file set ID of the file we are checking
+  # @param [String] file_id - The file ID that was checked
+  # @param [String] expected_result - The expected checksum
+  # @return [ChecksumAuditLog]
+  # @see .prune_history
+  def self.create_and_prune!(passed:, checked_uri:, file_set_id:, file_id:, expected_result:)
+    log = create!(passed: passed, checked_uri: checked_uri, file_set_id: file_set_id, file_id: file_id, expected_result: expected_result)
+    # A short-circuit. Given that .prune_history keeps all failing logs, there is no need to call prune history
+    # unless the check passed.
+    prune_history(file_set_id, checked_uri: checked_uri) if log.passed?
+    log
+  end
+
+  # Prune old ChecksumAuditLog records. We keep only:
+  # * Latest check
+  # * failing checks
+  # * any checks immediately before or after a failing check,
+  #   to provide context on known good dates surrounding failing.
+  def self.prune_history(file_set_id, checked_uri:)
+    all_logs = logs_for(file_set_id, checked_uri: checked_uri).reorder("created_at asc").to_a
+
+    0.upto(all_logs.length - 2).each do |i|
+      next if all_logs[i].failed?
+      next if i > 0 && all_logs[i - 1].failed?
+      next if all_logs[i + 1].failed?
+
+      all_logs[i].destroy!
+    end
+  end
+
+  # All logs for a particular file or version in a give file set, sorted
+  # by date descending.
+  def self.logs_for(file_set_id, checked_uri:)
+    ChecksumAuditLog.where(file_set_id: file_set_id, checked_uri: checked_uri).order('created_at desc, id desc')
   end
 end