Extracting single files from a tarball (spreadsheets)

Question

I've created a program that will pull spreadsheets out of a large tarball.

First it specifies the file name through the "folio" number which is given as an argument when the program is called. From there it runs a bash command that will list a text file containing the information needed to find the tarball, the information needed is a date. It then collects the information finds the tarball, runs another bash command to unarchive the spreadsheet from the tarball, and then moves the spreadsheets.

I would like some critique on my work. Are there better ways to extract from a tar? Would this be better as a module then a class?

I know it doesn't delete the directory that is created from extracting the spreadsheets, mostly because I don't trust myself enough to remove a directory called home.

#!/usr/local/bin/ruby

require 'rubygems'
require 'colored'
require 'fileutils'

class ArchiveManager

  attr_accessor :folio_num

  def initialize(folio_num)
    @folio_num = folio_num
    @src_dir = "/home/<dir>/<user>/<dir>/home/common/4000_UW_spreadsheets/#{folio_num}.ods"
    @dst_dir= "/home/<dir>/<user>/<dir>"
  end

  def num_check
    if folio_num =~ /^\d{4,7}$/
      search_for_spreadsheet
    else
      puts
      puts "                  Invalid Folio length".red.bold,
       "        USAGE: unarchive_ss <folio number 4-7 digits>".red.bold
      puts
    end
  end

  def search_for_spreadsheet
    location = FileUtils.chdir("/home/<dir>/<user>/<dir>")
    digits = `grep -lr #{@folio_num}.ods *.txt`
    file_name = digits.match(/\d+/).to_s
    puts "Spreadsheet found in UW_archive_#{file_name}.tgz".green.bold
    extract(file_name, location)
  end

  def extract(file_name, location)
    if File.exist?("#{@dst_dir}/#{@folio_num}.ods")
      found_error = raise SecurityError.new("File: #{@folio_num} exists within directory: #{@dst_dir}; will not continue..")
      abort("File exists in directory, either use file as parachute, or get rid of it..".red.bold)
    else
      tar_location = "home/common/4000_UW_spreadsheets"
      `tar -x -f UW_archive_#{file_name}.tgz #{tar_location}/#{@folio_num}.ods`
      move(tar_location, location)
    end
  end

  def move(tar_location, location)
    if File.exist?("#{@src_dir}")
      puts "File extracted, and being moved.".green.bold
      sleep(1)
      FileUtils.mv("#{@src_dir}", "#{@dst_dir}")
      puts "File: #{@folio_num}.ods, is now located in directory path: #{@dst_dir}".blue.bold
      puts "Remove created directories manually".red.bold
    else
      puts "Unable to locate file: #{@folio_num}.ods; source not found.".red.bold
      src_error = raise LoadError.new("Unable to find file: #{@folio_num}.ods; in directory #{@src_dir}")
    end
  end
end


ss = ArchiveManager.new(ARGV[0])
ss.num_check

Example of usage: File 2715015

Answer 1

Your main questions

Are there better ways to extract from a tar?

Using shell commands, no. Well, with any modern version of tar, instead of tar -x -f archive file you could write simpler as tar xf archive file.

I don't know Ruby, but probably there exist libraries to extract files from archives. It would be better to use that.

The current script is a mixture of Ruby and Bash commands. This is a fragile setup:

• It requires two technologies to run
• It requires understanding two technologies

The solution would be more robust if it was written either completely in Ruby or in Bash.

I know it doesn't delete the directory that is created from extracting the spreadsheets, mostly because I don't trust myself enough to remove a directory called home.

You could recursively remove the empty parent directories. That should be safe.

Naming

Looking at this usage:

ss = ArchiveManager.new(ARGV[0])
ss.num_check

This doesn't really tell what's going to happen. The program extracts a file from an archive. This would look like a more natural usage:

archive_manager = ArchiveManager.new()
archive_manager.extract_file(ARGV[0])

Design

The naming issue brought to the fore some design issues. The naming change I suggested earlier for more natural usage, also forces further changes that will lead to better design:

• Currently the ArchiveManager class is one-time use to extract a single file. That's not a natural way to use classes. Making the ArchiveManager reusable by adding a extract_file method with a file parameter is more natural.

• Currently many methods of ArchiveManager depend on the folio_num stored in the class. Methods are easier to understand when their dependencies are more directly visible, for example as input parameters. It also makes them easier to test.