# Counts the number of files in each subdirectory of the given directory
# and then outputs the count in a Tufte-esque style histogram, where the
# number before the | is the 10's column, and each number after the |
# represents one data point with that value. This lets you see the 
# spread to get a sense of the distribution. Each 10's value has two
# rows: one for values 0-4, one for 5-9
#
# 0|
# 0|99
# 1|0001233333333344444444444
# 1|55555555555556666666677777777777777777778888888888888888888888899999
# 2|0000000000001111111111111111112222222222222222333333333334444444444 
# 2|5555555555555555555666666667777788888888999
# 3|123 
# 3|
#

counts = []
dir = ARGV[0]
Dir.entries(dir).each do |entry|
  next if entry =~ /^\./ || !File.directory?("#{dir}/#{entry}")
  counts << Dir.entries("#{dir}/#{entry}").size - 2
end

display = Hash.new { |h,k| h[k] = [ [], [] ] }
counts.sort.each do |count|
  key = count / 10
  value = count % 10
  display[key][value < 5 ? 0 : 1] << value
end

display.keys.sort.each do |key|
  puts "#{key}|#{display[key][0].sort.join}"
  puts "#{key}|#{display[key][1].sort.join}"
end