This is a binary-to-text encoder which converts every 15 bits of input to a pair of printable characters in range U+0025..U+00ff. In other words, it generates 16 bytes of ISO8859-1-encoded bytes for every 15 bytes of binary data.

It was invented as a way to transmit binary files through serial devices using a delimiter-based, human-readable protocol. The output of Base32768 is free of whitespace (space, no-break space, soft hyphen etc.) and control characters. A few punctuation characters like + and = are also left out to be used as delimiters.

• High-efficiency: adds only 6.7% of overhead on average

• Small code and memory requirements (no lookup tables are needed)

• Uses only printable characters encodeable with ISO8859-1, meaning it's friendly for debugging and even text editors.

require 'base32768'

i = StringIO.new("\x17\x65\x23\x1c\x8c\xdc\xaa\xd9\xe5\xbd\x23\x91\xd7\x00\xa3")
o = StringIO.new
Base32768.encode(i, o)

File.open('out.txt', 'w+') { |f| f.write(o.string) }
=> 16
o.string.force_encoding('iso8859-1').encode('utf-8')
=> "õÙ¯w|ìf¸uЫ[ItR¾"

This project is intended to be a set of reference implementations for the encoder, with a single file per language.

To contribute:

• Fork the project

• Checkout a new branch

• Make a pull request

If you want to implement the encoder in a different language follow these rules:

• Use a single file

• Write some test-cases

• Make the program callable from the command line:

  • With -e it encodes the standard input to standard output

  • With -d it decodes the standard input to standard output

  • Otherwise it runs the test-cases

MIT. See LICENSE.