base64_spectrum_parser.rb

#!/usr/bin/ruby

require 'zlib'

NETWORK_FLOAT = 'g*'
NETWORK_DOUBLE = 'G*'
LITTLE_ENDIAN_FLOAT = 'e*'
LITTLE_ENDIAN_DOUBLE = 'E*'
BASE_64 = 'm*'

def unpack_mzxml_string(str, precision=32, network_order=true, compressed=false)
  unpack_code = 
    case precision
    when 32 ; network_order ?  NETWORK_FLOAT : LITTLE_ENDIAN_FLOAT
    when 64 ; network_order ?  NETWORK_DOUBLE : LITTLE_ENDIAN_DOUBLE
    else
      raise ArgumentError, "unknown precision (should be 32 or 64): #{precision}"
    end
  binary = str.unpack(BASE_64).first
  binary = Zlib::Inflate.inflate(binary) if compressed
  binary.unpack(unpack_code)
end


if $0 == __FILE__
  if ARGV.size == 0
    puts "usage: #{File.basename(__FILE__)} <base64string>"
    puts "outputs rows of: m/z<TAB>intensity"
    puts "currently assumes precision=32 and network order (mzXML default)"
    exit
  end

  str = ARGV.shift
  array_of_floats = unpack_mzxml_string(str)
  puts array_of_floats.each_slice(2).map {|pair| pair.join("\t") }.join("\n")
end