Skip to content

A faster alternative to base64 encoding and provides missing base encoding for OpenResty application

License

Notifications You must be signed in to change notification settings

spacewander/lua-resty-base-encoding

Repository files navigation

Name

lua-resty-base-encoding - Faster alternative to base64 encoding and provides missing base encoding for OpenResty application

All encoding are implemented in optimized C code with LuaJIT FFI binding.

Most of the inner encoding implementations are from Nick Galbreath's stringencoders. The base32 encoding is implemented by myself, but also inspired from his art work.

Build status: Travis

Table of Contents

MUST READ

  • The base64 encoding algorithm is ENDIAN DEPENDENT. The default version only works with little endian. To compile the big endian version, run make CEXTRAFLAGS="-DWORDS_BIGENDIAN" instead.
  • The base64 encoding algorithm assumes input string is ALIGNED, so it could be used only on x86(x64) and modern ARM architecture.

Synopsis

local base_encoding = require "resty.base_encoding"
local raw = "0123456789"

-- base32
local encoded = base_encoding.encode_base32(raw)
-- Or without '=' padding: local encoded = base_encoding.encode_base32(raw, true)
base_encoding.decode_base32(encoded) -- 0123456789

-- base64/base64_url (drop-in alternative to official API from lua-resty-core)
base_encoding.encode_base64(raw)
base_encoding.decode_base64(encoded)
base_encoding.encode_base64url(raw)
base_encoding.decode_base64url(encoded)

For more examples, read the t/base*.t files.

Installation

Run make. Then copy the librestybaseencoding.so to one of your lua_package_cpath. Yes, this library uses a trick to load the shared object from cpath instead of system shared library path. Finally, add the $pwd/lib to your lua_package_path.

Back to TOC

Benchmark

In short, it is quite faster than pure Lua alternatives, and a little faster than other C/FFI implementation. Go to Benchmark page for the details.

Back to TOC

Methods

encode_base2

syntax: encoded = encode_base2(raw)

Encode given string into base2 format(aka. bin format). Note that the input is string. Therefore, the encode_base2 result of 1 is 00110001, because the ascii value of 1 is 49, and the binary format of 49 is 00110001. And don't forget that the output of encode_base2 is a string instead of a binary number.

Back to TOC

decode_base2

syntax: raw, err = decode_base2(encoded)

Decode base2 format string into its raw value. If the given string is not valid base2 encoded, the raw will be nil and err will be "invalid input". Any character in the input string which is not 1 will be considered as 0. For example, aa11aaa1 is equal to 00110001. There is no RFC requires we should treat character not in 0 and 1 as invalid input, and check if a character is '0' or not will slow the performance down by 50%.

Back to TOC

encode_base16

syntax: encoded = encode_base16(raw[, out_in_lowercase])

Encode given string into base16 format(aka. hex/hexadecimal format). This method may be named to_hex or encodeHex in other languages. The default output letters are in [0-9A-F]. If you specify the out_in_lowercase to true, the output will be in [0-9a-f].

Back to TOC

decode_base16

syntax: raw, err = decode_base16(encoded)

Decode base16 format(aka. hex/hexadecimal format) string into its raw value. This method may be named from_hex or decodeHex in other languages. If the given string is not valid base16 encoded, the raw will be nil and err will be "invalid input". Letters in [0-9a-fA-F] are considered valid.

Back to TOC

encode_base32

syntax: encoded = encode_base32(raw[, no_padding])

Encode given string into base32 format with/without padding '='. The default value of no_padding is false.

Back to TOC

decode_base32

syntax: raw, err = decode_base32(encoded)

Decode base32 format string into its raw value. If the given string is not valid base32 encoded, the raw will be nil and err will be "invalid input".

Back to TOC

encode_base32hex

syntax: encoded = encode_base32hex(raw[, no_padding])

Encode given string into base32hex format with/without padding '='. The default value of no_padding is false. For more info of base32hex format, see https://tools.ietf.org/html/rfc4648#section-7.

Back to TOC

decode_base32hex

syntax: raw, err = decode_base32(encoded)

Decode base32hex format string into its raw value. If the given string is not valid base32hex encoded, the raw will be nil and err will be "invalid input". For more info of base32hex format, see https://tools.ietf.org/html/rfc4648#section-7.

Back to TOC

encode_base64

decode_base64

encode_base64url

decode_base64url

Drop-in alternative to the official implementation in lua-resty-core. Read their official documentation instead. The encode method is 40% faster, and the decode method is 200% faster. Note that the implementation is endian and architecture dependent. Read the 'Must Read' section for more info.

Back to TOC

encode_base85

syntax: encoded = encode_base85(raw)

Encode given string into base85 format with/without padding '='. Note that there is not a standard but too many variants of so-called base85. This module's implementation should be compatiable with Go's encoding/ascii85 module (not in the level of API argument, but in the level of encode/decode rules).

Back to TOC

decode_base85

syntax: raw, err = decode_base85(encoded)

Decode base85 format string into its raw value. If the given string is not valid base85 encoded, the raw will be nil and err will be "invalid input".

Back to TOC