Ruby script to download GitHub repositories, tar & gzip and upload to S3.

repository_archiver.rb

#!/usr/bin/env ruby

require 'optparse'
require 'aws-sdk'

options = {}
opt_parser = OptionParser.new do |opt|
  opt.banner = "Usage: AWS_ACCESS_KEY_ID=123 AWS_SECRET_ACCESS_KEY=123 repository_archiver.rb --archive unused-fork,some_project,rails"

  opt.on("-a","--archive unused-fork,some_project,rails","Name of repositories to clone & archive to S3",Array) do |repositories|
    options[:repositories] = repositories
  end

  opt.on("-h","--help","help") do
    abort(opt_parser.banner)
  end
end

opt_parser.parse!

abort(opt_parser.banner) if options[:repositories].nil?


def clone repository
  puts "\nCloning '#{repository}'"
  return puts "Already cloned" if File.directory? repository
  url = "[email protected]:lonelyplanet/#{repository}.git"
  `git clone #{url}`
  abort "\nFailed to clone '#{url}'" unless $?.success?
end

def archive repository
  puts "\nArchiving '#{repository}'"
  target = "_archived_#{repository}.tgz"
  `tar -czf #{target} #{repository}`
  return target
end

def upload tarball_name
  puts "\nUploading '#{tarball_name}'"
  obj = @s3.buckets['archived-repositories'].objects[tarball_name]
  obj.write(file: tarball_name)
end

AWS.eager_autoload! AWS::S3
@s3 = AWS::S3.new(region: 'us-east-1')

options[:repositories].each do |repository|
  puts "\n----------------------------\n"
  clone(repository)
  upload archive(repository)
end