unix_utils
Like FileUtils, but provides zip, unzip, bzip2, bunzip2, tar, untar, sed, du, md5sum, shasum, cut, head, tail, wc, unix2dos, dos2unix, iconv, curl, perl, etc.
You must have these binaries in your PATH
. Not a pure-ruby implementation of all these UNIX greats!
Works in MRI 1.8.7+, MRI 1.9.2+, and JRuby 1.6.7+. No gem dependencies; uses stdlib
Real-world usage
We use unix_utils
for data science at Brighter Planet and in production at
Originally extracted from remote_table
Philosophy
Use a subprocess to perform a big task and then get out of memory.
Rules (what you can expect)
For commands like zip, untar, sed, head, cut, dos2unix, etc.:
- Just returns a path to the output, randomly named, located in the system tmp dir (
UnixUtils.unzip('kittens.zip)
→'/tmp/unix_utils-129392301-kittens'
) - Never touches the input
- Sticks a useful file extension on the output, if applicable (
UnixUtils.tar('puppies/')
→'/tmp/unix_utils-99293192-puppies.tar'
)
For commands like du, md5sum, shasum, etc.:
- Just returns the good stuff (the checksum, for example, not the filename that is listed after it in the standard command output)
- Never touches the input
But I can just spawn these myself
This lib was created to ease the pain of remembering command options for Gentoo, deciding which spawning method to use, possibly handling pipes...
require 'tmpdir'
destdir = File.join(Dir.tmpdir, "kittens_#{Kernel.rand(1e11)}")
require 'open3'
Open3.popen3('unzip', '-q', '-n', 'kittens.zip, '-d', destdir) do |stdin, stdout, stderr|
stdin.close
@error_message = stderr.read
end
is replaced safely with
destdir = UnixUtils.unzip 'kittens.zip'
Digest::SHA256
But I can just use (Note: Balazs Kutil pointed out this is a bad example... I will replace it soon)
This will load an entire file into memory before it can be processed...
require 'digest'
str = Digest::SHA256.hexdigest File.read('kittens.zip')
... so you're really replacing this ...
sha256 = Digest::SHA256.new
File.open('kittens.zip', 'r') do |f|
while chunk = f.read(4_194_304)
sha256 << chunk
end
end
str = sha256.hexdigest
You get the same low memory footprint with
str = UnixUtils.shasum 'kittens.zip', 256
Compatibility
Uses open3
because it's in the Ruby stdlib and is consistent across MRI and JRuby.
Wishlist
- cheat sheet based on GNU Coreutils cheat sheet
- yarddocs
- properly use Dir.tmpdir(name), etc.
- smarter tmp file name generation - don't include url params for curl, etc.
Authors
- Seamus Abshere [email protected]
Copyright
Copyright (c) 2012 Seamus Abshere