Skip to content
Efficient and customizable data-encoding functions in Rust
Branch: master
Clone or download
Latest commit cda5148 Apr 22, 2019
Type Name Latest commit message Commit time
Failed to load latest commit information.
bin Address clippy warnings Jan 1, 2019
cmp cmp: Bump lazy_static Apr 22, 2019
lib Release macro 0.1.7 Apr 22, 2019
Makefile lib/macro: Use proc-macro-hack for stable Nov 18, 2017
appveyor.yml www: Improvements Jul 5, 2018


Latest Version Documentation Latest License Build Status Build Status Coverage Status

This repository provides a Rust library and a binary providing efficient common and custom data-encodings.

Common use-cases

The library provides the following common encodings:

  • HEXLOWER: lowercase hexadecimal
  • HEXLOWER_PERMISSIVE: lowercase hexadecimal with case-insensitive decoding
  • HEXUPPER: uppercase hexadecimal
  • HEXUPPER_PERMISSIVE: uppercase hexadecimal with case-insensitive decoding
  • BASE32: RFC4648 base32
  • BASE32_NOPAD: RFC4648 base32 without padding
  • BASE32_DNSSEC: RFC5155 base32
  • BASE32_DNSCURVE: DNSCurve base32
  • BASE32HEX: RFC4648 base32hex
  • BASE32HEX_NOPAD: RFC4648 base32hex without padding
  • BASE64: RFC4648 base64
  • BASE64_NOPAD: RFC4648 base64 without padding
  • BASE64_MIME: RFC2045-like base64
  • BASE64URL: RFC4648 base64url
  • BASE64URL_NOPAD: RFC4648 base64url without padding

Typical usage looks like:

// allocating functions
// in-place functions
BASE32.encode_mut(&input_to_encode, &mut encoded_output)
BASE64_URL.decode_mut(&input_to_decode, &mut decoded_output)

See the documentation or the changelog for more details.

Custom use-cases

The library also provides the possibility to define custom little-endian ASCII base-conversion encodings for bases of size 2, 4, 8, 16, 32, and 64 (for which all above use-cases are particular instances). It supports:

  • padded and unpadded encodings
  • canonical encodings (e.g. trailing bits are checked)
  • in-place encoding and decoding functions
  • partial decoding functions (e.g. for error recovery)
  • character translation (e.g. for case-insensitivity)
  • most and least significant bit-order
  • ignoring characters when decoding (e.g. for skipping newlines)
  • wrapping the output when encoding

The typical definition of a custom encoding looks like:

lazy_static! {
    static ref HEX: Encoding = {
        let mut spec = Specification::new();
    static ref BASE64: Encoding = {
        let mut spec = Specification::new();
        spec.padding = Some('=');

You may also use the macro library to define a compile-time custom encoding:

const HEX: Encoding = new_encoding!{
    symbols: "0123456789abcdef",
    translate_from: "ABCDEF",
    translate_to: "abcdef",
const BASE64: Encoding = new_encoding!{
    symbols: "ABCDEFGHIJKLMNOPQRSTUVWXYZabcdefghijklmnopqrstuvwxyz0123456789+/",
    padding: '=',

See the documentation or the changelog for more details.


The performance of the encoding and decoding functions (for both common and custom encodings) are similar to existing implementations in C, Rust, and other high-performance languages. You may run the benchmarks with make bench.

Swiss-knife binary

The binary is mostly a wrapper around the library. You can run make install to install it from the repository. By default, it will be installed as ~/.cargo/bin/data-encoding. You can also run cargo install data-encoding-bin to install the latest version published on This second alternative does not require to clone the repository.

Once installed, you can run data-encoding --help (assuming ~/.cargo/bin is in your PATH environment variable) to see the usage:

Usage: data-encoding --mode=<mode> --base=<base> [<options>]
Usage: data-encoding --mode=<mode> --symbols=<symbols> [<options>]

    -m, --mode <mode>   {encode|decode|describe}
    -b, --base <base>   {16|hex|32|32hex|64|64url}
    -i, --input <file>  read from <file> instead of standard input
    -o, --output <file> write to <file> instead of standard output
        --block <size>  read blocks of about <size> bytes
    -p, --padding <padding>
                        pad with <padding>
    -g, --ignore <ignore>
                        when decoding, ignore characters in <ignore>
    -w, --width <cols>  when encoding, wrap every <cols> characters
    -s, --separator <separator>
                        when encoding, wrap with <separator>
        --symbols <symbols>
                        define a custom base using <symbols>
        --translate <new><old>
                        when decoding, translate <new> as <old>
                        when decoding, ignore non-zero trailing bits
                        use least significant bit first bit-order

    # Encode using the RFC4648 base64 encoding
    data-encoding -mencode -b64     # without padding
    data-encoding -mencode -b64 -p= # with padding

    # Encode using the MIME base64 encoding
    data-encoding -mencode -b64 -p= -w76 -s$'\r\n'

    # Show base information for the permissive hexadecimal encoding
    data-encoding --mode=describe --base=hex

    # Decode using the DNSCurve base32 encoding
    data-encoding -mdecode \
        --symbols=0123456789bcdfghjklmnpqrstuvwxyz \
        --translate=BCDFGHJKLMNPQRSTUVWXYZbcdfghjklmnpqrstuvwxyz \
You can’t perform that action at this time.