#!/usr/bin/ruby

# == Synopsis
#
# Output a report of where your hard disk space has gone, organized by folder
# (subdirectory), with the most bloated folders at the top.
#
# == Usage
#
#   bloat [ -h | --help ] [ [ -b | --binary ] | [ -d | --decimal ] ] dir
#
# dir:: 
#   The starting directory. Disk space is summarized for all directories 
#   below it.
#
# --binary reports bloat in base 2 units (MiB, GiB)
#
# --decimal reports bloat in base 10 units (MB, GB), the default because
#           that's how hard disks are measured [1]
#
# --files includes sizes for all files, not just directories
#
# [1] http://lpar.ath0.com/2008/07/15/si-unit-prefixes-a-plea-for-sanity/
#
# == Author
# mathew (mailto:meta@pobox.com)
#
# == Copyright
# Copyright © 2005 mathew. Licensed under the same terms as Ruby.

require 'getoptlong'
require 'find'
require 'rubygems'
require 'rdoc'

# Default @base for output
@base = 10

@files = false

# Return a number in appropriate units, nicely justified, as a string
def formattednumber(x)
  u = 0
  while (x > @divisor)
    x = x / @divisor
    u = u + 1
  end
  return sprintf("%4d %-#{@width}s", x, @suffix[u])
end

# Now the main program begins

# Get arguments
opts = GetoptLong.new(
  [ '--help', '-h', GetoptLong::NO_ARGUMENT ],
  [ '--binary', '-b', GetoptLong::NO_ARGUMENT ],
  [ '--decimal', '-d', GetoptLong::NO_ARGUMENT ],
  [ '--files', '-f', GetoptLong::NO_ARGUMENT ]
)

# Parse them and act on them
opts.each do |opt, arg|
  case opt
    when '--help'
      # used to do RDoc::usage, but that was removed from rdoc and I haven't
      # found a good replacement yet
    when '--binary'
      @base = 2
    when '--decimal'
      @base = 10
    when '--files'
      @files = true;
    else
      puts "Unrecognized option #{opt}"
      exit 0
  end
end

# Now work out the unit suffixes and ratio between them
if @base == 2
  @suffix = ['B', 'KiB', 'MiB', 'GiB', 'TiB']
  @divisor = 1024
  @width = 3
else
  @suffix = ['B', 'kB', 'MB', 'GB', 'TB']
  @divisor = 1000
  @width = 2
end

# Should be 1 remaining argument, the directory to start with
if ARGV.length != 1 
  #RDoc::usage
  exit 0
end

# Perform the find and total the bloat
bloat = Hash.new
Find.find(ARGV[0]) do |file|
  # Find.prune if file == '.'
  stat = File.lstat(file)
  dir = File.dirname(file)
  if stat.file?
    if bloat[dir]
      bloat[dir] += stat.size
    else
      bloat[dir] = stat.size
    end
  end
  if @files
    bloat[file] = stat.size
  end
end

# Sort by reverse bloatyness
bloatarray = bloat.sort {|pair1, pair2| pair2[1] <=> pair1[1]}

# Output
for pair in bloatarray
  print formattednumber(pair[1]), ' ', pair[0], "\n"
end