add infrastructure for clearing repository and running seeds

.dassie/.env

@@ -1,3 +1,4 @@
 MEMCACHED_HOST=memcached
 REDIS_PROVIDER=SIDEKIQ_REDIS_URL
-SIDEKIQ_REDIS_URL=redis://redis:6379/0
+SIDEKIQ_REDIS_URL=redis://redis:6379/0
+SEED_DASSIE=true

.dassie/db/seeds.rb

@@ -1,22 +1,77 @@
-# This file should contain all the record creation needed to seed the database with its default values.
-# The data can then be loaded with the rails db:seed command (or created alongside the database with db:setup).
-#
-# Examples:
-#
-#   movies = Movie.create([{ name: 'Star Wars' }, { name: 'Lord of the Rings' }])
-#   Character.create(name: 'Luke', movie: movies.first)
+wipe_data = ActiveModel::Type::Boolean.new.cast(ENV.fetch('WIPE_DATA', false))
+seed_release_testing = ActiveModel::Type::Boolean.new.cast(ENV.fetch('SEED_RELEASE_TESTING', false))
+seed_dassie = ActiveModel::Type::Boolean.new.cast(ENV.fetch('SEED_DASSIE', false))
 
-Hyrax::Engine.load_seed unless ActiveModel::Type::Boolean.new.cast(ENV["SKIP_HYRAX_ENGINE_SEED"])
+unless wipe_data || seed_release_testing
+  puts 'NAME'
+  puts '     rails db:seed (Hyrax)'
+  puts
+  puts 'SYNOPSIS'
+  puts '     bundle exec rails db:seed [WIPE_DATA=true|false] [SEED_RELEASE_TESTING=true|false] [SEED_DASSIE=true|false]'
+  puts
+  puts 'DESCRIPTION'
+  puts '     Hyrax defined db:seed provides a means to clear repository metadata from the datastore (e.g. Fedora, Postgres) and from Solr.'
+  puts '     Seeds can be run to pre-populate metadata to help with release testing and local development testing.'
+  puts
+  puts '     NOTE: Options can be passed in with the command on the command line or set as ENV variables.'
+  puts
+  puts '     The options are as follows:'
+  puts
+  puts '     WIPE_DATA'
+  puts '             USE WITH CAUTION - Deleted data cannot be recovered.'
+  puts
+  puts '             When true, it will clear all repository metadata from the datastore (e.g. Fedora, Postgres) and from Solr.  It also'
+  puts '             clears data from the application database that are tightly coupled to repository metadata.  See Hyrax::DataMaintenance'
+  puts '             for more information on what data will be destroyed by this process.'
+  puts
+  puts '             The wipe_data process will also restore required repository metadata including collection types and the default admin'
+  puts '             set.  See Hyrax::RequiredDataSeeder for more information on what data will be created by this process.'
+  puts
+  puts '     SEED_RELEASE_TESTING'
+  puts '             When true, it will run the set of seeds for release testing creating a repository metadata and support data, including'
+  puts '             test users, collection types, collections, and works with and without files.  See Hyrax::TestDataSeeder for more information'
+  puts '             on what data will be created by this process.'
+  puts
+  puts '     SEED_DASSIE'
+  puts '             When true, it will run a minimal set of seeds for dassie test app, including required collection types, default admin set,'
+  puts '             and test users.'
+  puts
+  puts '     ALLOW_RELEASE_SEEDING_IN_PRODUCTION'
+  puts '             USE WITH EXTERME CAUTION WHEN USED IN PRODUCTION - Deleted data cannot be recovered.  Attempts are made to not overwrite'
+  puts '             existing data, but use in production is not recommended.'
+  puts
+  puts '             If this is NOT true, the process will abort when Rails environment is production.'
+  puts
+end
+
+allow_release_seeding_in_production = ActiveModel::Type::Boolean.new.cast(ENV.fetch('ALLOW_RELEASE_SEEDING_IN_PRODUCTION', false))
+
+if Rails.env == 'production' && !allow_release_seeding_in_production
+  puts "Seeding data for release testing is not for use in production!"
+  exit
+end
+
+if wipe_data
+  puts '####################################################################################'
+  puts
+  puts 'WARNING: You are about to clear all repository metadata from the datastore and solr.'
+  puts 'Are you sure? [YES|n]'
+  answer = STDIN.gets.chomp
+  unless answer == 'YES'
+    puts '   Aborting!'
+    puts '####################################################################################'
+    exit
+  end
 
-puts "\n== Loading users"
-User.where(email: 'admin@example.com').first_or_create do |f|
-  f.password = 'admin_password'
+  Hyrax::DataMaintenance.new.destroy_repository_metadata_and_related_data
+  Hyrax::RequiredDataSeeder.new.generate_seed_data
 end
 
-User.where(email: 'basic_user@example.com').first_or_create do |f|
-  f.password = 'password'
+if seed_dassie
+  Hyrax::RequiredDataSeeder.new.generate_seed_data
+  Hyrax::TestDataSeeders::UserSeeder.generate_seeds
 end
 
-User.where(email: 'another_user@example.com').first_or_create do |f|
-  f.password = 'password'
+if seed_release_testing
+  Hyrax::TestDataSeeder.new.generate_seed_data
 end

app/utils/hyrax/data_destroyers/collection_branding_destroyer.rb

@@ -0,0 +1,29 @@
+# frozen_string_literal: true
+module Hyrax
+  module DataDestroyers
+    # Collection branding info is tightly coupled to collections.  When they are
+    # removed using wipe!, the associated database entries for collection branding
+    # also have to be deleted.
+    #
+    # @note WARNING: DO NOT USE IN PRODUCTION!  The methods in this class are destructive.
+    #   Data can not be recovered.
+    #
+    # @todo This destroys branding info in the database.  Should it also delete
+    #   related banner and logo files?
+    class CollectionBrandingDestroyer
+      class << self
+        attr_accessor :logger
+
+        def destroy_data(logger: Logger.new(STDOUT), allow_destruction_in_production: false)
+          raise("CollectionBrandingDestroyer is not for use in production!") if Rails.env.production? && !allow_destruction_in_production
+          @logger = logger
+
+          logger.info("Destroying collection branding...")
+
+          Hyrax::CollectionBrandingInfo.destroy_all
+          logger.info("   collection branding -- DESTROYED")
+        end
+      end
+    end
+  end
+end

app/utils/hyrax/data_destroyers/collection_types_destroyer.rb

@@ -0,0 +1,26 @@
+# frozen_string_literal: true
+module Hyrax
+  module DataDestroyers
+    # Collection types are recreated by the release seed data.  Clear out here to
+    # start with a fresh set.
+    #
+    # @note WARNING: DO NOT USE IN PRODUCTION!  The methods in this class are destructive.
+    #   Data can not be recovered.
+    #
+    class CollectionTypesDestroyer
+      class << self
+        attr_accessor :logger
+
+        def destroy_data(logger: Logger.new(STDOUT), allow_destruction_in_production: false)
+          raise("CollectionTypesDestroyer is not for use in production!") if Rails.env.production? && !allow_destruction_in_production
+          @logger = logger
+
+          logger.info("Destroying collection types...")
+
+          Hyrax::CollectionType.destroy_all
+          logger.info("   collection types -- DESTROYED")
+        end
+      end
+    end
+  end
+end

app/utils/hyrax/data_destroyers/default_admin_set_id_cache_destroyer.rb

@@ -0,0 +1,26 @@
+# frozen_string_literal: true
+module Hyrax
+  module DataDestroyers
+    # When the default admin set is removed using wipe!, the cache of the default
+    # admin set id also needs to be deleted.
+    #
+    # @note WARNING: DO NOT USE IN PRODUCTION!  The methods in this class are destructive.
+    #   Data can not be recovered.
+    #
+    class DefaultAdminSetIdCacheDestroyer
+      class << self
+        attr_accessor :logger
+
+        def destroy_data(logger: Logger.new(STDOUT), allow_destruction_in_production: false)
+          raise("DefaultAdminSetIdCacheDestroyer is not for use in production!") if Rails.env.production? && !allow_destruction_in_production
+          @logger = logger
+
+          logger.info("Destroying default admin set id cache...")
+
+          Hyrax::DefaultAdministrativeSet.destroy_all
+          logger.info("   default admin set id cache -- DESTROYED")
+        end
+      end
+    end
+  end
+end

app/utils/hyrax/data_destroyers/featured_works_destroyer.rb

@@ -0,0 +1,27 @@
+# frozen_string_literal: true
+module Hyrax
+  module DataDestroyers
+    # Featured works are tightly coupled to works in the repository. When they
+    # are removed using wipe!, the associated database entries for featured
+    # works also need to be deleted.
+    #
+    # @note WARNING: DO NOT USE IN PRODUCTION!  The methods in this class are destructive.
+    #   Data can not be recovered.
+    #
+    class FeaturedWorksDestroyer
+      class << self
+        attr_accessor :logger
+
+        def destroy_data(logger: Logger.new(STDOUT), allow_destruction_in_production: false)
+          raise("FeaturedWorksDestroyer is not for use in production!") if Rails.env.production? && !allow_destruction_in_production
+          @logger = logger
+
+          logger.info("Destroying featured works...")
+
+          FeaturedWork.destroy_all
+          logger.info("   featured works -- DESTROYED")
+        end
+      end
+    end
+  end
+end

app/utils/hyrax/data_destroyers/permission_templates_destroyer.rb

@@ -0,0 +1,30 @@
+# frozen_string_literal: true
+module Hyrax
+  module DataDestroyers
+    # PermissionTemplates are tightly coupled to admin sets and collections.
+    # When they are removed using wipe!, the associated database entries for
+    # permission templates also have to be deleted.
+    #
+    # @note WARNING: DO NOT USE IN PRODUCTION!  The methods in this class are destructive.
+    #   Data can not be recovered.
+    #
+    class PermissionTemplatesDestroyer
+      class << self
+        attr_accessor :logger
+
+        def destroy_data(logger: Logger.new(STDOUT), allow_destruction_in_production: false)
+          raise("PermissionTemplatesDestroyer is not for use in production!") if Rails.env.production? && !allow_destruction_in_production
+          @logger = logger
+
+          logger.info("Destroying permission templates...")
+
+          Hyrax::PermissionTemplateAccess.destroy_all
+          logger.info("   permission templates access -- DESTROYED")
+
+          Hyrax::PermissionTemplate.destroy_all
+          logger.info("   permission templates -- DESTROYED")
+        end
+      end
+    end
+  end
+end

app/utils/hyrax/data_destroyers/repository_metadata_destroyer.rb

@@ -0,0 +1,42 @@
+# frozen_string_literal: true
+module Hyrax
+  module DataDestroyers
+    # Clear all repository metadata from the datastore (e.g. Fedora, Postgres)
+    # and from Solr.
+    #
+    # @note WARNING: DO NOT USE IN PRODUCTION!  The methods in this class are destructive.
+    #   Data can not be recovered.
+    #
+    # @note When using Wings adapter, which wraps ActiveFedora, the persister wipe!
+    #   method clears both Fedora and Solr.  Optionally, an index_adapter may also
+    #   be set to use Valkyrie to write to a second Solr core.  If the Hyrax index
+    #   adapter is the NullIndexingAdapter, that means that Valkyrie is not being
+    #   used to index the repository metadata and does not need to be cleared using
+    #   the Hyrax index adapter.
+    #
+    #   When using other Valkyrie persistence adapters, wipe! only clears the repository
+    #   metadata from the datastore (e.g. ORM table in Postgres).  The index adapter
+    #   is used to clear metadata from Solr.
+    #
+    class RepositoryMetadataDestroyer
+      class << self
+        attr_accessor :logger
+
+        def destroy_metadata(logger: Logger.new(STDOUT), allow_destruction_in_production: false)
+          raise("RepositoryMetadataDestroyer is not for use in production!") if Rails.env.production? && !allow_destruction_in_production
+          @logger = logger
+
+          logger.info("Destroying all repository metadata...")
+
+          unless Hyrax.index_adapter.is_a? Valkyrie::Indexing::NullIndexingAdapter
+            conn = Hyrax.index_adapter.connection
+            conn.delete_by_query('*:*', params: { 'softCommit' => true })
+          end
+          Hyrax.persister.wipe!
+
+          logger.info("   repository metadata -- DESTROYED")
+        end
+      end
+    end
+  end
+end

app/utils/hyrax/data_destroyers/stats_destroyer.rb

@@ -0,0 +1,33 @@
+# frozen_string_literal: true
+module Hyrax
+  module DataDestroyers
+    # Stats are tightly coupled to works and files in the repository. When they
+    # are removed using wipe!, the associated database entries for stats also
+    # need to be deleted.
+    #
+    # @note WARNING: DO NOT USE IN PRODUCTION!  The methods in this class are destructive.
+    #   Data can not be recovered.
+    #
+    class StatsDestroyer
+      class << self
+        attr_accessor :logger
+
+        def destroy_data(logger: Logger.new(STDOUT), allow_destruction_in_production: false)
+          raise("StatsDataDestroyer is not for use in production!") if Rails.env.production? && !allow_destruction_in_production
+          @logger = logger
+
+          logger.info("Destroying stats...")
+
+          FileDownloadStat.destroy_all
+          logger.info("   file download stats -- DESTROYED")
+
+          FileViewStat.destroy_all
+          logger.info("   file view stats -- DESTROYED")
+
+          WorkViewStat.destroy_all
+          logger.info("   work view stats -- DESTROYED")
+        end
+      end
+    end
+  end
+end

app/utils/hyrax/data_maintenance.rb

@@ -0,0 +1,51 @@
+# frozen_string_literal: true
+
+module Hyrax
+  ##
+  # These methods created for use in rake tasks and db/seeds.rb  They can be used to:
+  # * clear repository metadata and related data and files
+  # * clear temporary files
+  # * clear logs
+  #
+  # @note WARNING: DO NOT USE IN PRODUCTION!  The methods in this class are destructive.
+  #   Data can not be recovered.
+  #
+  class DataMaintenance
+    attr_accessor :logger, :allow_destruction_in_production
+
+    def initialize(logger: Logger.new(STDOUT), allow_destruction_in_production: false)
+      raise("Destruction of data is not for use in production!") if Rails.env.production? && !allow_destruction_in_production
+      @logger = logger
+      @allow_destruction_in_production = allow_destruction_in_production
+    end
+
+    # Clear repository metadata and related data
+    # * clear repository metadata from the datastore (e.g. Fedora, Postgres) and from Solr
+    # * clear targeted application data that is tightly coupled to repository metadata
+    # * delete files that are tightly coupled to repository metadata
+    def destroy_repository_metadata_and_related_data
+      Hyrax::DataDestroyers::RepositoryMetadataDestroyer.destroy_metadata(logger: logger, allow_destruction_in_production: allow_destruction_in_production)
+      Hyrax::DataDestroyers::StatsDestroyer.destroy_data(logger: logger, allow_destruction_in_production: allow_destruction_in_production)
+      Hyrax::DataDestroyers::FeaturedWorksDestroyer.destroy_data(logger: logger, allow_destruction_in_production: allow_destruction_in_production)
+      Hyrax::DataDestroyers::PermissionTemplatesDestroyer.destroy_data(logger: logger, allow_destruction_in_production: allow_destruction_in_production)
+      Hyrax::DataDestroyers::CollectionBrandingDestroyer.destroy_data(logger: logger, allow_destruction_in_production: allow_destruction_in_production)
+      Hyrax::DataDestroyers::DefaultAdminSetIdCacheDestroyer.destroy_data(logger: logger, allow_destruction_in_production: allow_destruction_in_production)
+      Hyrax::DataDestroyers::CollectionTypesDestroyer.destroy_data(logger: logger, allow_destruction_in_production: allow_destruction_in_production)
+
+      # TODO: Stubbed until RepositoryFilesDestroyer is written
+      # Hyrax::DataDestroyers::RepositoryFilesDestroyer.destroy_data(logger: logger, allow_destruction_in_production: allow_destruction_in_production)
+    end
+
+    # @todo write code to clear out log files
+    def destroy_log_files
+      # Stubbed until LogFilesDestroyer is written
+      # Hyrax::DataDestroyers::LogFilesDestroyer.destroy_data(logger: logger, allow_destruction_in_production: allow_destruction_in_production)
+    end
+
+    # @todo write code to delete tmp files
+    def destroy_tmp_files
+      # Stubbed until TmpFilesDestroyer is written
+      # Hyrax::DataDestroyers::TmpFilesDestroyer.destroy_data(logger: logger, allow_destruction_in_production: allow_destruction_in_production)
+    end
+  end
+end

app/utils/hyrax/required_data_seeder.rb

@@ -0,0 +1,21 @@
+# frozen_string_literal: true
+module Hyrax
+  # This class was created for use in rake tasks and db/seeds.rb.  It generates
+  # required repository metadata including Admin Set and User collection types and
+  # the default admin set.
+  #
+  # Seeders of required data are non-destructive.  If the data already exists,
+  # it will not be replaced.
+  class RequiredDataSeeder
+    attr_accessor :logger
+
+    def initialize(logger: Logger.new(STDOUT))
+      @logger = logger
+    end
+
+    def generate_seed_data
+      Hyrax::RequiredDataSeeders::CollectionTypeSeeder.generate_seeds(logger: logger)
+      Hyrax::RequiredDataSeeders::CollectionSeeder.generate_seeds(logger: logger)
+    end
+  end
+end