From: GOTO Kentaro <gotoken@...>
Date: 2002-07-20T03:02:52+09:00
Subject: Re: Can I know a file is binary or text in ruby?

At Sat, 20 Jul 2002 01:20:56 +0900,
mookhae wrote:
> I am newbie in ruby and computer language.
> To make simple ruby script, I need a file information.
> Which file is binary or text.
> In ruby, how can I know file information?

It is impossible to determine completely whether a file is binary or
not because binary file is undefined concept.  Unix command `file'
does that by heuristics.  file command is pragmatically enough well
but not complete. 

Now, the following tests if a file includes non ascii printable code
point byte.  You can improve this script to detect non latin-1 etc.
However some coding systems is stateful, for example iso-2022, and
this approach does not work for such character coding systems. 

  #! ruby
  NON_ASCII_PRINTABLE = /[^\x20-\x7e\s]/

  def nonbinary?(io, forbidden, size = 1024)
    while buf = io.read(size)
      return false if forbidden =~ buf
    end
    true
  end

  # usage: ruby this_script.rb filename ...
  ARGV.each do |fn|
    begin
      open(fn) do |f|
        if nonbinary?(f, NON_ASCII_PRINTABLE)
          puts "#{fn}: ascii printable"
        else
          puts "#{fn}: binary"
        end
      end
    rescue
      puts "#$0: #$!"
    end
  end