Merge pull request #4881 from sul-dlss/audit-migration

Audit migration

bin/audit-repository-migration

@@ -0,0 +1,56 @@
+#!/usr/bin/env ruby
+# frozen_string_literal: true
+
+# Aduits migration of DRO/Collection/AdminPolicy to RepositoryObject.
+
+require_relative '../config/environment'
+require 'tty-progressbar'
+
+# how many druids to process as a single advance unit of the progress bar
+def num_for_progress_advance(count)
+  return 1 if count < 100
+
+  count / 100
+end
+
+def on_finish(results, progress_bar)
+  progress_bar.advance(results.size)
+end
+
+def tty_progress_bar(count, model)
+  TTY::ProgressBar.new(
+    "#{model} [:bar] (:percent (:current/:total), rate: :rate/s, mean rate: :mean_rate/s, :elapsed total, ETA: :eta_time)",
+    bar_format: :crate,
+    advance: num_for_progress_advance(count),
+    total: count
+  )
+end
+
+def audit_one(legacy_model)
+  model = RepositoryObject.find_by(external_identifier: legacy_model.external_identifier)
+  [legacy_model.class.name, legacy_model.external_identifier] unless model && legacy_model.version == model.versions.count && Cocina::Models.without_metadata(legacy_model.to_cocina) == Cocina::Models.without_metadata(model.head_version.to_cocina)
+end
+
+def audit_all(ar_class:)
+  druids = ar_class.pluck(:external_identifier)
+
+  progress_bar = tty_progress_bar(druids.length, ar_class.name)
+  progress_bar.start
+
+  Parallel.map(druids.each_slice(100),
+               in_processes: 4,
+               finish: ->(_, _, results) { on_finish(results, progress_bar) }) do |druids_slice|
+    druids_slice.map do |druid|
+      audit_one(ar_class.find_by(external_identifier: druid))
+    end
+  end.flatten(1).compact
+end
+
+output = [Dro, Collection, AdminPolicy].each_with_object([]) do |klass, log|
+  log.concat(audit_all(ar_class: klass))
+end
+
+puts 'Class,External ID'
+output.each do |row|
+  puts row.join(',')
+end