Compare list of files in Globus with list of files in PDC Describe

file-compare.rb

require "json"

doi = ARGV[0]
if doi == nil
  puts "Must provide a DOI"
  return
end

# List of files from Globus
# fetch via https://github.com/pulibrary/rdss-handbook/blob/main/globus_cli.md#a-complete-example
txt_file = "#{doi}.txt"
txt_files = File.readlines(txt_file).map(&:chomp)

# List of files from PDC Describe
# https://datacommons.princeton.edu/describe/works/467/file-list?_=1729628275940
json_file = "#{doi}.json"
json_files = JSON.parse(File.read(json_file))

puts "Globus count: #{txt_files.count}, PDC count: #{json_files.count}"
puts "Extra files:"
if json_files.count > txt_files.count
  json_files.each do |file|
    filename = file["filename_display"]
    if txt_files.find {|x| x == filename} != nil
      # puts "OK - #{filename}"
    else
      puts "#{filename}"
      # require "byebug"; byebug
    end
  end
end