Adds Fix for deleting PhysicalInstantiations from Assets based on Med…

…ia Type (#922)

Refactors some one-type scripts into more re-usable batch processing methods that can be run
from the ruby console or CLI.

* Delete Assets by ID list from file.
* Delete Physical Instantiations by Media Type and IDs from file.
* Change Media Type for instantiations by IDs from file.

lib/fix/batch_process.rb

@@ -0,0 +1,89 @@
+module Fix
+  class BatchProcess
+    attr_reader :ids, :log, :cli_ptions, :log_level
+
+    def initialize(ids_file:, log_level: Logger::INFO)
+      @ids = File.readlines(ids_file, chomp: true)
+      @cli_options = {}
+      @log = Logger.new(STDOUT)
+      @log.level = log_level
+    end
+
+    # asset_resources Returns an array of AssetResource instances for the given ids.
+    # @return [Array<AssetResource>] An array of AssetResource instances.
+    def asset_resources
+      @asset_resources ||= ids.map do |id|
+        log.info "Finding Asset Resource '#{id}'..."
+        begin
+          AssetResource.find(id)
+        rescue => e
+          log_error(e)
+          nil
+        end
+      end.compact
+    end
+
+    def log_error(e)
+      log.error "#{e.class}: #{e.message}"
+      log.debug "Backtrace:\n#{e.backtrace.join("\t\t\n")}\n\n"
+    end
+
+    # run! is the main method to be implemented by subclasses.
+    def run
+      log.warn "No action taken. Put your logic in the #{self.class}#run! method"
+    end
+
+    # self.cli_options A hash to store command line options.
+    def self.cli_options
+      @cli_options ||= {}
+    end
+
+    # self.option_parser Creates a default OptionParser for cli options and allows subclasses
+    # to add their own options.
+    # @param block [Proc] A block that takes an OptionParser instance as an argument.
+    # @return [OptionParser] The OptionParser instance.
+    # Usage:
+    #  class MyBatchProcess < BatchProcess
+    #    def initialize(my_option:, **args)
+    #      super(**args)
+    #      @my_option = my_option
+    #    end
+    #
+    #    option_parser do |opts|
+    #      opts.on("-m", "--my-option", "My custom option") do |my_option_val|
+    #        # Assign option values to the cli_options hash.
+    #        cli_options[:my_option] = my_option_val
+    #      end
+    #    end
+    #  end
+    def self.option_parser(&block)
+      # Set a default options for all BatchProcess classes
+      @option_parser ||= OptionParser.new do |opts|
+        # Allow verbose ouput
+        opts.on('-l', '--log-level [0-5]', '0=DEBUG, 1=INFO, 2=WARN, 3=ERROR, 4=FATAL, 5=UNKNOWN') do |log_level|
+          cli_options[:log_level] = log_level[/\d+/].to_i || 1
+        end
+
+        # Allow file input of AAPB IDs
+        opts.on("-f", "--file FILE", "List of AAPB IDs, one per line") do |file|
+          cli_options[:ids_file] = file
+        end
+      end
+
+      # Call the passed block with option parser instance if a block was given.
+      block.call(@option_parser) if block_given?
+
+      # Return the option parser.
+      @option_parser
+    end
+
+    # self.run_cli Parses command line options and runs the batch process.
+    def self.run_cli
+      # Call option_parser.parse! to set cli_options from $ARGV
+      option_parser.parse!
+
+      # Run the batch process with cli_options
+      new(**cli_options).run
+    end
+  end
+end

lib/fix/change_media_type.rb

@@ -1,53 +1,56 @@
 require_relative '../../config/environment'
+require_relative 'batch_process'
 
 module Fix
-  class ChangeMediaType
-    MEDIA_TYPES = MediaTypeService.new.select_all_options
+  class ChangeMediaType < BatchProcess
+    MEDIA_TYPES = MediaTypeService.new.select_all_options.to_h.values
 
-    attr_reader :aapb_ids_filename, :aapb_ids
-    def initialize
-      @aapb_ids_filename = "#{File.dirname(File.expand_path(__FILE__))}/data/nebraska_public_media_ids.txt"
+    attr_reader :media_type
+
+    def initialize(media_type:, **args)
+      super(**args)
+      raise ArgumentError, "media_type must be one of '#{MEDIA_TYPES.join("', '")}', but '#{media_type}' was given" unless MEDIA_TYPES.include?(media_type)
+      @media_type = media_type
     end
+
+    option_parser do |opts|
+      opts.banner = "Usage: ruby lib/fix/change_media_type.rb [options]"
 
-    def aapb_ids
-      @aapb_ids ||= File.readlines(aapb_ids_filename).map(&:strip)
+      opts.on("-t", "--media-type MEDIA_TYPE", "Either 'Sound' or 'Movind Image'") do |media_type|
+        cli_options[:media_type] = media_type
+      end
     end
 
-    def asset_resources
-      @asset_resources ||= aapb_ids.map do |aapb_id|
-        puts "Looking up Asset #{aapb_id} ..."
-        begin
-          AssetResource.find(aapb_id)
-        rescue => e
-          puts "Error looking up Asset #{aapb_id}: #{e.class} -- #{e.message}"
+    def run
+      log.info "Running fix #{self.class.name} ..."
+      asset_resources.each do |ar|
+        if ar.physical_instantiation_resources.count == 0
+          log.warn "No Physical Instantiations for Asset Resource #{ar.id}, skipping."
+          next
         end
-      end.compact
-    end
 
-    def run!
-      puts "Running Hotfix #{self.class.name} ..."
-      asset_resources.each do |asset_resource|
-        pi = asset_resource.physical_instantiation_resources.detect  { |pi| pi.media_type != 'Moving Image' }
+        pi = ar.physical_instantiation_resources.detect  { |pi| pi.media_type != media_type }
         if !pi
-          puts "Nothing to fix for AssetResource #{asset_resource.id}, skipping ..."
+          log.warn "Asset Resource #{ar.id} has no Physical Instantiations without media type of #{media_type}, skipping."
           next
         end
 
         # Change the metadata
-        pi.media_type = 'Moving Image'
+        pi.media_type = media_type
 
         begin
           pi.save
-          puts "PhysicalInstantiationResource #{pi.id} saved with media_type 'Moving Image'"
+          log.info "Physical Instantiation #{pi.id} for Asset Resource #{ar.id} saved with media_type '#{media_type}'"
         rescue => e
-          puts "Error saving PhysicalInstantiationResource #{pi.id}: #{e.class} -- #{e.message}"
+          log_error e
         end
       end
+      log.info "Done."
     end
   end
 end
 
 
 if __FILE__ == $0
-  Fix::ChangeMediaType.new.run!
+  Fix::ChangeMediaType.run_cli
 end

lib/fix/delete_asset_resources.rb

@@ -0,0 +1,25 @@
+require_relative '../../config/environment'
+require_relative 'batch_process'
+
+module Fix
+  class DeleteAssetResources < BatchProcess
+    def run
+      asset_resources.each do |ar|
+        log.info "Destroying Asset Resource #{ar.id}"
+        begin
+          Hyrax.persister.delete(resource: ar)
+          Hyrax.index_adapter.delete(resource: ar)
+          Hyrax.index_adapter.connection.commit
+          log.info "Asset Resource #{ar.id} destroyed."
+        rescue => e
+          log_error e
+        end
+      end
+      puts "Done."
+    end
+  end
+end
+
+if __FILE__ == $0
+  Fix::DeleteAssetResources.run_cli
+end