sizes: Calculate and sort all filesizes for current folder

sizes.rb

#!/usr/bin/env ruby
# Sizes - Calculate and sort all filesizes for current folder Includes
# directory sizes, colorized output Brett Terpstra 2019 WTF License
VERSION = "1.0.1"

require 'shellwords'

# Just including term-ansicolor by @flori and avoiding all the
# rigamarole of requiring multiple files when it's not a gem... - Brett
#
# ansicolor Copyright: Florian Frank
# License: <https://github.com/flori/term-ansicolor/blob/master/COPYING>
# Home: <https://github.com/flori/term-ansicolor>
module Term

  # The ANSIColor module can be used for namespacing and mixed into your own
  # classes.
  module ANSIColor
    # require 'term/ansicolor/version'

    # :stopdoc:
    ATTRIBUTES = [
      [ :clear              ,   0 ],     # String#clear is already used to empty string in Ruby 1.9
      [ :reset              ,   0 ],     # synonym for :clear
      [ :bold               ,   1 ],
      [ :dark               ,   2 ],
      [ :italic             ,   3 ],     # not widely implemented
      [ :underline          ,   4 ],
      [ :underscore         ,   4 ],     # synonym for :underline
      [ :blink              ,   5 ],
      [ :rapid_blink        ,   6 ],     # not widely implemented
      [ :negative           ,   7 ],     # no reverse because of String#reverse
      [ :concealed          ,   8 ],
      [ :strikethrough      ,   9 ],     # not widely implemented
      [ :black              ,  30 ],
      [ :red                ,  31 ],
      [ :green              ,  32 ],
      [ :yellow             ,  33 ],
      [ :blue               ,  34 ],
      [ :magenta            ,  35 ],
      [ :cyan               ,  36 ],
      [ :white              ,  37 ],
      [ :on_black           ,  40 ],
      [ :on_red             ,  41 ],
      [ :on_green           ,  42 ],
      [ :on_yellow          ,  43 ],
      [ :on_blue            ,  44 ],
      [ :on_magenta         ,  45 ],
      [ :on_cyan            ,  46 ],
      [ :on_white           ,  47 ],
      [ :intense_black      ,  90 ],    # High intensity, aixterm (works in OS X)
      [ :intense_red        ,  91 ],
      [ :intense_green      ,  92 ],
      [ :intense_yellow     ,  93 ],
      [ :intense_blue       ,  94 ],
      [ :intense_magenta    ,  95 ],
      [ :intense_cyan       ,  96 ],
      [ :intense_white      ,  97 ],
      [ :on_intense_black   , 100 ],    # High intensity background, aixterm (works in OS X)
      [ :on_intense_red     , 101 ],
      [ :on_intense_green   , 102 ],
      [ :on_intense_yellow  , 103 ],
      [ :on_intense_blue    , 104 ],
      [ :on_intense_magenta , 105 ],
      [ :on_intense_cyan    , 106 ],
      [ :on_intense_white   , 107 ]
    ]

    ATTRIBUTE_NAMES = ATTRIBUTES.transpose.first
    # :startdoc:

    # Returns true if Term::ANSIColor supports the +feature+.
    #
    # The feature :clear, that is mixing the clear color attribute into String,
    # is only supported on ruby implementations, that do *not* already
    # implement the String#clear method. It's better to use the reset color
    # attribute instead.
    def support?(feature)
      case feature
      when :clear
        !String.instance_methods(false).map(&:to_sym).include?(:clear)
      end
    end
    # Returns true, if the coloring function of this module
    # is switched on, false otherwise.
    def self.coloring?
      @coloring
    end

    # Turns the coloring on or off globally, so you can easily do
    # this for example:
    #  Term::ANSIColor::coloring = STDOUT.isatty
    def self.coloring=(val)
      @coloring = val
    end
    self.coloring = true

    ATTRIBUTES.each do |c, v|
      eval <<-EOT
        def #{c}(string = nil)
          result = ''
          result << "\e[#{v}m" if Term::ANSIColor.coloring?
          if block_given?
            result << yield
          elsif string.respond_to?(:to_str)
            result << string.to_str
          elsif respond_to?(:to_str)
            result << to_str
          else
            return result #only switch on
          end
          result << "\e[0m" if Term::ANSIColor.coloring?
          result
        end
      EOT
    end

    # Regular expression that is used to scan for ANSI-sequences while
    # uncoloring strings.
    COLORED_REGEXP = /\e\[(?:(?:[349]|10)[0-7]|[0-9])?m/

    # Returns an uncolored version of the string, that is all
    # ANSI-sequences are stripped from the string.
    def uncolored(string = nil) # :yields:
      if block_given?
        yield.to_str.gsub(COLORED_REGEXP, '')
      elsif string.respond_to?(:to_str)
        string.to_str.gsub(COLORED_REGEXP, '')
      elsif respond_to?(:to_str)
        to_str.gsub(COLORED_REGEXP, '')
      else
        ''
      end
    end

    module_function

    # Returns an array of all Term::ANSIColor attributes as symbols.
    def attributes
      ATTRIBUTE_NAMES
    end
    extend self
  end
end

# Begin sizes

class String

  include Term::ANSIColor

  # ensure trailing slash
  def slashit
    self.sub(/\/?$/,'/')
  end

  # colorize a human readable size format by size
  def color_fmt
    case self
    when /\dB?$/
      self.blue
    when /\dKB?$/
      self.green
    when /\dMB?$/
      self.yellow
    when /\dGB?$/
      self.red
    else
      self.bold.red
    end
  end

  # colorize files by type (directories and hidden files)
  def color_file(force_check=false)
    filename = self.dup
    if force_check && File.directory?(filename)
      filename.sub!(/\/?$/,'/')
    end

    case filename
    when /\/$/
      filename.green
    when /^\./
      filename.white
    else
      filename.bold.white
    end
  end

  # Replace $HOME in path with ~
  def short_dir
    home = ENV['HOME']
    self.sub(/#{home}/, '~')
  end

  # Convert a line like `120414 filename` to a colorized string with
  # human readable size
  def line_to_human
    parts = self.split(/\t/)
    if parts[0] =~ /NO ACCESS/
      "  ERROR".red + " " + parts[1].color_file
    else
      size = to_human(parts[0].to_i).color_fmt
      size.pad_escaped(7) + " " + parts[1].color_file
    end
  end

  # Pad a line containing ansi escape codes to a given length, ignoring
  # the escape codes
  def pad_escaped(len)
    str = self.dup
    str.gsub!(/\e\[\d+m/,'')
    prefix = ""
    while prefix.length + str.length < len
      prefix += " "
    end
    prefix + self
  end
end

# Convert a number (assumed bytes) to a human readable format (12.5K)
def to_human(n,fmt=false)
  count = 0
  formats = %w(B K M G T P E Z Y)

  while  (fmt || n >= 1024) && count < 8
    n /= 1024.0
    count += 1
    break if fmt && formats[count][0].upcase =~ /#{fmt[0].upcase}/
  end

  format("%.2f%s",n,formats[count])
end

# Use `du` to size a single directory and all of its contents. This
# number is returned in blocks (512B), so the human readable result may
# be slightly different than you'd get from `ls` or a GUI file manager
def du_size_single(dir)
  res = %x{du -s #{Shellwords.escape(dir)} 2>/dev/null}.strip
  if $?.success?
    parts = res.split(/\t/)
    (parts[0].to_i * 512).to_s + "\t" + parts[1].strip
  else
    "NO ACCESS\t#{dir}"
  end
end

# main function
def all_sizes(dir)
  # Use `ls` to list all files in the target with long info
  files = %x{ls -lSrAF "#{dir.slashit}" 2>/dev/null}
  unless $?.success?
    $stdout.puts "Error getting file listing".red
    Process.exit 1
  end
  files = files.strip.split(/\n/)

  files.delete_if {|line|
    line.strip =~ /^total \d+/
  }

  # trim file list to just size and filename
  files.map! {|line|
    line.sub(/\S{10,11} +\d+ +\S+ +\w+ +(\d+) +\w{3} +\d+ +[\d:]+ +(.*?)$/, "\\1\t\\2")
  }


  # if a line is a path to a directory, use `du` to update its size with
  # the total filesize of the directory contents.
  files.map! {|entry|
    file = entry.split(/\t/)[1]
    if File.directory?(file)
      du_size_single(file)
    else
      entry
    end
  }

  # Sort by size (after updating directory sizes)
  files.sort! {|a,b|
    size1 = a.split(/\t/)[0].to_i
    size2 = b.split(/\t/)[0].to_i
    size1 <=> size2
  }

  # Output each line with human-readable size and colorization
  files.each {|entry|
    $stdout.puts entry.line_to_human
  }
  # Include a total for the directory
  $stdout.puts "-------".black.bold
  $stdout.puts(du_size_single(dir).short_dir.line_to_human)
end

def help
  app = File.basename(__FILE__)
  help =<<EOHELP

#{app.bold.white} #{VERSION.green} by Brett Terpstra
  Display a human-readable list of sizes for all files and directories.

usage:
  $ #{app.bold.white} [directory]

Leaving directory blank operates in the current working directory.

EOHELP
  puts help
  Process.exit 0
end

# Assume operating on current directory...
dir = ENV['PWD']

# ...unless an argument is provided
if ARGV[0]
  # Add some help. Why not?
  if ARGV[0] =~ /^-?-h(elp)?/
    help
  elsif ARGV[0] =~ /^-?-v(ersion)?/
    $stdout.puts File.basename(__FILE__) + " v" + VERSION
    Process.exit 0
  else
    argdir = File.expand_path(ARGV[0])
    if File.directory?(argdir)
      dir = argdir
    end
  end
end

all_sizes(dir)

I added double quotes here to allow for spaces in the top directory (otherwise getting 'error getting file sizes message')

files = %x{ls -lSrAF "#{dir.slashit}" 2>/dev/null}

It would be good to get a total sum of the files it could size rather than just error out because of eg .Trashes/ being unavailable to measure.

@rleaver152 thanks, updated the gist.

Line 264 seems to fail when there is a - in a username. This makes sense since - is not part of w+ in regex.

I changed it locally from
line.sub(/\S{10,11} +\d+ +\S+ +\w+ +(\d+) +\w{3} +\d+ +[\d:]+ +(.*?)$/, "\\1\t\\2")

to
line.sub(/\S{10,11} +\d+ +\S+ +[\w-]+ +(\d+) +\w{3} +\d+ +[\d:]+ +(.*?)$/, "\\1\t\\2")

I think that will also include dashes in the search for user/group now. I don't know if it breaks anything else though 😉

Line 264 seems to fail when there is a - in a username. This makes sense since - is not part of w+ in regex.

I changed it locally from
line.sub(/\S{10,11} +\d+ +\S+ +\w+ +(\d+) +\w{3} +\d+ +[\d:]+ +(.*?)$/, "\\1\t\\2")

to
line.sub(/\S{10,11} +\d+ +\S+ +[\w-]+ +(\d+) +\w{3} +\d+ +[\d:]+ +(.*?)$/, "\\1\t\\2")

I think that will also include dashes in the search for user/group now. I don't know if it breaks anything else though 😉

Could also just use \S to get all non-whitespace characters, just in case... Not sure offhand what all is valid in usernames, though.