cobs2

Crates.iocobs2
lib.rscobs2
version0.1.4
sourcesrc
created_at2024-06-20 11:47:21.377399
updated_at2024-06-23 23:53:44.741127
descriptionConsistent Overhead Byte Stuffing — COBS — and variant COBS/R
homepage
repositoryhttps://github.com/cmcqueen/cobs2-rs
max_upload_size
id1277892
size109,104
Craig McQueen (cmcqueen)

documentation

README

Consistent Overhead Byte Stuffing (COBS)

Rust functions for encoding and decoding COBS.

Intro

This crate provides functions for encoding and decoding of COBS, and a minor variant of COBS known as COBS/R.

What Is COBS?

COBS is a method of encoding a packet of bytes into a form that contains no bytes with value zero (0x00). The input packet of bytes can contain bytes in the full range of 0x00 to 0xFF. The COBS encoded packet is guaranteed to generate packets with bytes only in the range 0x01 to 0xFF. Thus, in a communication protocol, packet boundaries can be reliably delimited with 0x00 bytes.

The COBS encoding does have to increase the packet size to achieve this encoding. However, compared to other byte-stuffing methods, the packet size increase is reasonable and predictable. COBS always adds 1 byte to the message length. Additionally, for longer packets of length n, it may add n/254 (rounded down) additional bytes to the encoded packet size.

For example, compare to the PPP protocol, which uses 0x7E bytes to delimit PPP packets. The PPP protocol uses an "escape" style of byte stuffing, replacing all occurences of 0x7E bytes in the packet with 0x7D 0x5E. But that byte-stuffing method can potentially double the size of the packet in the worst case. COBS uses a different method for byte-stuffing, which has a much more reasonable worst-case overhead.

For more details about COBS, see the references.

I have included a variant on COBS, COBS/R, which slightly modifies COBS to often avoid the +1 byte overhead of COBS. So in many cases, especially for smaller packets, the size of a COBS/R encoded packet is the same size as the original packet. See below for more details about COBS/R.

References

Consistent Overhead Byte Stuffing
Stuart Cheshire and Mary Baker
IEEE/ACM Transations on Networking, Vol. 7, No. 2, April 1999

Consistent Overhead Byte Stuffing (for IEEE)
http://www.stuartcheshire.org/papers/COBSforToN.pdf

PPP Consistent Overhead Byte Stuffing (COBS)
PPP Working Group Internet Draft
James Carlson, IronBridge Networks
Stuart Cheshire and Mary Baker, Stanford University
November 1997

PPP Consistent Overhead Byte Stuffing (COBS)
http://tools.ietf.org/html/draft-ietf-pppext-cobs-00

Modules Provided

  • cobs2::cobs — Consistent Overhead Byte Stuffing (basic method)
  • cobs2::cobsr — COBS/R — Consistent Overhead Byte Stuffing—Reduced

"Consistent Overhead Byte Stuffing—Reduced" (COBS/R) is my own invention, a modification of basic COBS encoding, and is described in more detail below.

The following are not implemented:

  • COBS/ZPE — Consistent Overhead Byte Stuffing—Zero Pair Elimination
  • COBS/ZRE — Consistent Overhead Byte Stuffing—Zero Run Elimination

Usage

The modules provide functions for encoding and decoding. Several implementations are provided, which differ in the input and output data types.

  • Arrays (no_std)
    • encode_array()
    • decode_array()
  • Vectors
    • encode_vector()
    • decode_vector()
  • Iterators
    • encode_iter()
    • decode_iter()
    • encode_ref_iter()
    • decode_ref_iter()

Unit Testing

Unit testing is implemented:

cargo test

License

The code is released under the MIT license. See LICENSE.txt for details.

Consistent Overhead Byte Stuffing—Reduced (COBS/R)

A modification of COBS, which I'm calling "Consistent Overhead Byte Stuffing—Reduced" (COBS/R), is provided in the cobs2::cobsr module. Its purpose is to save one byte from the encoded form in some cases. Plain COBS encoding always has a +1 byte encoding overhead. See the references for details. COBS/R can often avoid the +1 byte, which can be a useful savings if it is mostly small messages that are being encoded.

In plain COBS, the last length code byte in the message has some inherent redundancy: if it is greater than the number of remaining bytes, this is detected as an error.

In COBS/R, instead we opportunistically replace the final length code byte with the final data byte, whenever the value of the final data byte is greater than or equal to what the final length value would normally be. This variation can be unambiguously decoded: the decoder notices that the length code is greater than the number of remaining bytes.

Examples

The byte values in the examples are in hex.

First example

Input:

2F A2 00 92 73 02

This example is encoded the same in COBS and COBS/R. Encoded (length code bytes are highlighted):

03 2F A2 04 92 73 02

Second example

The second example is almost the same, except the final data byte value is greater than what the length byte would be.

Input:

2F A2 00 92 73 26

Encoded in plain COBS (length code bytes are highlighted):

03 2F A2 04 92 73 26

Encoded in COBS/R:

03 2F A2 26 92 73

Because the last data byte (26) is greater than the usual length code (04), the last data byte can be inserted in place of the length code, and removed from the end of the sequence. This avoids the usual +1 byte overhead of the COBS encoding.

The decoder detects this variation on the encoding simply by detecting that the length code is greater than the number of remaining bytes. That situation would be a decoding error in regular COBS, but in COBS/R it is used to save one byte in the encoded message.

Commit count: 76

cargo fmt