NAME

basenc, base64, base64url, base32, base32hex, base16, base2msbf, base2lsbf, z85 — transcode RFC 4648, binary, or ZeroMQ data

SYNOPSIS

DESCRIPTION

Without -d, encode file (standard input stream if "-", the default), mapping consecutive bits to the indices into one of the following RFC 4648 alphabets:

base2?sbf use a different algorithm:

z85 uses the ZeroMQ 32/Z85 algorithm, mapping consecutive 4-byte chunks onto 5 output bytes by casting them to big-endian 32-bit unsigned integers, then repeatedly dividing by 85, with the remainders used as an index into the alphabet (output in reverse order):

If wrap (default 76) is non-zero, newlines are inserted every wrap bytes, as well as a final one if the output wasn't empty.

With -d, decode it, mapping input bytes from the alphabet to their indices' bits. Padding, if any, is required to decode the full message.

base2?sbf decode in the expected way. z85 does the inverse operation, treating each 5 bytes as a base-85 big-endian number, and outputting the resulting integer in big-endian order. The encoded input not being a multiple of 8 or 5 is a hard error.

If bytes from outside the alphabet (except the newline) are encountered and -i wasn't specified, a diagnostic is issued.

By definition, sequentially applying either two inverse transformations yields the same data: encoding into any of these alphabets, carefully composed of bytes which universally have no special meaning, allows lossless transmission of binary data as plain text at only a minor increase in size, equal to the amount of alphabet text required to express one byte of input: $\frac{8}{{\log }_2\left(\mathrm{len}\right(\mathrm{alphabet}\left)\right)}$ ⇒ 4/3=1.(3), 1.6, 2, 8 and 1.25 respectively.

base16 is equivalent to a hexadecimal listing and base2msbf to a binary one. base32hex has the useful property of retaining the sort ordering of the input (i.e. base32hex(data) < base32hex(dat2) ⇔ data < dat2). The base64url alphabet is safe to use in paths (and URLs), and may be converted into from the base64 alphabet by a simple tr '+/' '-_' invocation.

OPTIONS

-d, --decode

Decode the input.

-i, --ignore-garbage

Don't produce diagnostics and return successfully when a non-alphabet byte is encountered while decoding.

-w, --wrap=wrap

Wrap the encoded output at wrap columns and terminate it with a newline. Defaults to 76.

basenc requires a --algorithm flag to select the algorithm, but is otherwise equivalent to invoking algorithm directly with the same arguments.

EXIT STATUS

1 if file couldn't be opened or read, contained garbage and -i wasn't specified, or didn't contain a multiple of 4 bytes in z85 mode.

SEE ALSO

RFC 4648
ZeroMQ 32/Z85

STANDARDS

Compatible with the GNU system, which only contains base64, base32, and basenc. A compatible base64 also appears in NetBSD 9.0.

HISTORY

4BSD introduced uuencode(1C) (and uudecode(1C)): a structured, whole-file approach including the name and permissions in the encoded output, also using a 6-bits-per-byte encoding (but a fundamentally different one), an enhancement to the uucp(1)/uusend(1) suite. IEEE Std 1003.1-2001 (“POSIX.1”) added -m, using the basenc encoding instead.

coreutils 6.1 (2006-08-19) adds base64, largely as present-day. base32 joins it in coreutils 8.25 (2016-01-20), and basenc in coreutils 8.31 (2019-03-10).