Okay this got a bit more RFC than necessary but it should be unambiguous

Base2

Encoding

A byte array (or a string represented by a byte array, eg. UTF-8) is encoded to multibase base2 by prefixing the character 0 to the encoding of the byte array in base2. The length of the multibase base2 encoding of a byte array of length k is 8k+1 bytes.

The encoding of the byte array in base2 is obtained by concatenating the base2 representations of each byte in the array, in the order the bytes are in the array.

The base2 representation of a byte is a byte array of length 8 in big-endian order, where each byte of the array is set to the character 1, if the corresponding bit in the byte is set, and the character 0 if the corresponding bit is unset.

The character 0 means the hexadecimal value 0x30 (ASCII character 0, decimal value 48).

The character 1 means the hexadecimal value 0x31 (ASCII character 1, decimal value 49).

For example, the base2 representation of the ASCII character Y, or the hexadecimal value 0x59 is "01011001". The corresponding multibase base2 encoding of the character Y is "001011001"

The base2 encoding of any byte array must not contain any other bytes apart from the characters 0 and 1.

The entire process of encoding the byte array [0x58, 0x59, 0x60] is as follows:

map each byte to the base2 representation:
  ["01011000", "01011001", "01011010"]

concatenate:
  "010110000101100101011010"

prefix with '0':
  "0010110000101100101011010"

Decoding

The canonical format of multibase base2 encoding is decoded by dropping the first byte (which must be the character 0), dividing the remaining byte array to k sections of length 8, and mapping to each byte in an output buffer of k bytes the binary value of the corresponding section.

The binary value of a byte section is obtained by taking the bytes of the section in big-endian order, and setting the corresponding bit in the output byte if the byte is 1, and leaving the bit unset if 0.

For example, the binary value of the byte section "00110111" is 0x37.

The process, with input "00100000101000010":

drop the leading zero:
  "0100000101000010"
divide to sections:
  ["01000001", "01000010"]
allocate output buffer:
  [0x00, 0x00]
set the correct bits:
  [0x41, 0x42] or "AB"

A non-canonical encoding of multibase base2 whose length is not equal to 8k+1 for some k, may be decoded by dropping the first byte, and then prefixing to the encoding the smallest number of bytes that makes the length of the encoding be canonical, and then decoding it as if it were in the canonical base2 form. For example, the non-canonical encoding "011010" may be decoded like the canonical base2 encoding "000" + "11010" to get 0x1a. A standard multibase implementation must not produce non-canonical encoding.

test vectors #24

Description

Metadata

Metadata

Assignees

Labels

Type

Projects

Milestone

Relationships

Development

Issue actions