Quick script to shuffle a directory full of files into a test and training set

sample.rb

require 'fileutils'

# Given a directory of files, sort them randomly then split into a training and
# test set

TEST_PERCENT = 20

if ARGV.size < 1
  puts "Usage: #{$0} <path> (outdir)"
  exit 0
end

srcdir, outdir = *ARGV
outdir ||= '.'

# Get all the files in the specified src dir
files = Dir.entries(srcdir).collect do |file|
  path = File.join(srcdir, file)
  path if File::file?(path)
end

# Compact, uniq and shuffle!
files = files.compact.uniq.shuffle

# Now split into two sets
test_files = files[0, files.size * (TEST_PERCENT / 100.0)]
training_files = files - test_files

puts "Test files: #{test_files.inspect}"
puts "Training files: #{training_files.inspect}"

def migrate_files(files, category, outdir)
  files.each do |old_file|
    new_file = File.join outdir, category, File.basename(old_file)
    FileUtils.makedirs File.dirname(new_file)
    FileUtils.cp old_file, new_file
  end
end

migrate_files test_files, 'testing', outdir
migrate_files training_files, 'training', outdir