Ascii85 Encoder and Decoder

What is Ascii85?

Ascii85 reads the input 4 bytes at a time, treats each group as a 32-bit number and writes it as 5 base-85 digits. The digits are the 85 consecutive ASCII characters from ! (code 33) to u (code 117): each character is simply the digit value plus 33. The output is 25% larger than the input, against 33% for Base64.

The format was first used by the Unix btoa utility. Adobe adopted it as the ASCII base-85 encoding of PostScript Level 2 and PDF, where it lets binary data such as images and fonts travel inside a plain text file. Python offers it as base64.a85encode() and Go as the encoding/ascii85 package.

The z shortcut and the Adobe delimiters

A group of four zero bytes would normally be written as five exclamation marks. Ascii85 writes it as a single z instead, which keeps runs of empty data short. The shortcut applies only to complete groups. When the input length is not a multiple of 4, the final group of n bytes becomes n + 1 characters and is never abbreviated.

PostScript marks an Ascii85 string with <~ at the start and ~> at the end. In a PDF stream with the ASCII85Decode filter, only the closing ~> is used, as an end-of-data marker. The decoder here removes either delimiter when present and ignores whitespace, so you can paste stream data from a PDF or PostScript file directly.

Ascii85 in code

# Python 3.4+
import base64
base64.a85encode(b'hello')                      # b'BOu!rDZ'
base64.a85encode(b'hello', adobe=True)          # b'<~BOu!rDZ~>'
base64.a85decode(b'<~BOu!rDZ~>', adobe=True)    # b'hello'

// Go
import "encoding/ascii85"
dst := make([]byte, ascii85.MaxEncodedLen(len(src)))
n := ascii85.Encode(dst, src)