package osbx

  1. Overview
  2. Docs
Implementation of SeqBox in OCaml


Dune Dependency






A single file container/archive that can be reconstructed even after total loss of file system structures (from official SeqBox).

Published: 09 Aug 2017



Implementation of SeqBox in OCaml

The original pharsing was "Port of SeqBox to OCaml", but since no direct porting/translation was actually done due to the differences between Python 3(implementation language of SeqBox) and OCaml, which also force the architecture and design of this software(ocaml-SeqBox) to be independently developed, thus this is only an implementation(not a port or translation) of SeqBox according to its technical specifications. This is mainly to address the different licenses being used(SeqBox was using AGPL 3.0 at the time of writing while this project uses 3-Clause BSD license).

Official SeqBox Repo -

Table of Contents

Created by gh-md-toc


I would like to thank Marco (author of the official SeqBox) for discussing and clarifying several aspects of his project, and providing me with test data

I would like to thank Ming for his feedback on the documentation, UX design, and several other general aspects of this project. And also his help on testing the building and installation of osbx on macOS.

Getting started

Osbx 1.1.1 is currently on OPAM, 1.2.0 is currently in the process of being published

opam install osbx

There are four modes/commands for osbx currently : encode, decode, rescue, show

You can consult the man pages of osbx itself and all the four commands via

osbx        --help
osbx encode --help
osbx decode --help
osbx rescue --help
osbx show   --help


CRC-CCITT is currently implemented in pure OCaml and is translated from implementation in libcrc

  • See src/, src/crcccitt.mli for the OCaml implementation

    • The translated source code is under the same MIT license used by and stated in libcrc source code

  • See libcrc_crcccitt/crcccitt.c, libcrc_crcccitt/checksum.h for the source code from libcrc used for the translation

Exact behaviours in non-standard cases are not specified in official SeqBox technical specification

  • See specification of ocaml-SeqBox for details on how ocaml-SeqBox behaves(if you care about undefined behaviour those sort of things)

Hashing libraries

  • Nocrypto for SHA1, SHA256, SHA512

  • Digestif for BLAKE2B_512


Enabling silent mode is likely to speed things up

  • This is due to progress text printing does quite a lot of formatting, and checking time interval has some overhead



Index of source code



Possibly useful additional features of ocaml-SeqBox(possibly not yet in official SeqBox)

  • Allows random ordering in sbx container

    • This also means block corruption will not stop the decoding process

  • Allows duplicate metadata/data blocks to exist within one sbx container

    • This means you can concatenate multiple copies of sbx container together directly to increase chance of recovery in case of corruption

Technical Specification

The following specification is copied directly from the official specification (with possible slight modifications).

Also see section "Features currently NOT planned to be implemented" for features ocaml-SeqBox is probably not going to have

Byte order: Big Endian

Common blocks header:

pos to pos size desc
0 2 3 Recoverable Block signature = 'SBx'
3 3 1 Version byte
4 5 2 CRC-16-CCITT of the rest of the block (Version is used as starting value)
6 11 6 file UID
12 15 4 Block sequence number

Block 0

pos to pos size desc
16 n var encoded metadata
n+1 blockend var padding (0x1a)

Blocks > 0 & < last:

pos to pos size desc
16 blockend var data

Blocks == last:

pos to pos size desc
16 n var data
n+1 blockend var padding (0x1a)


N.B. Current versions differs only by blocksize.

ver blocksize note
1 512 default
2 128
3 4096

Metadata encoding:

Bytes Field
3 ID
1 Len
n Data
ID Desc
FNM filename (utf-8)
SNM sbx filename (utf-8)
FSZ filesize (8 bytes)
FDT date & time (8 bytes, seconds since epoch)
SDT sbx date & time (8 bytes)
HSH crypto hash (using Multihash protocol)
PID parent UID (not used at the moment)

Supported crypto hashes in 1.1.0 are

  • SHA1

  • SHA256

  • SHA512

  • BLAKE2B_512

Features currently NOT planned to be implemented
  • Data hiding (XOR encoding/decoding in official seqbox)

    • Provides neither sufficiently strong encryption nor sufficient stealth for any serious attempt to hide/secure data

    • You should use the appropriate tools for encryption


The following files directly from libcrc are under the MIT License(see license text in each of the code files)

  • libcrc_crcccitt/crcccitt.c

  • libcrc_crcccitt/checksum.h

The following files translated from libcrc source code are under the same MIT License as used by libcrc and as stated in libcrc source code, the license text of the crcccitt.c is copied over to src/ as well

  • src/

  • src/crcccitt.mli

The files in tests folder copied from official SeqBox are under its license, which is AGPL as of time of writing

  • tests/SeqBox/*

All remaining files are distributed under the 3-Clause BSD license as stated in the LICENSE file

Dependencies (8)

  1. cmdliner build
  2. hex build
  3. angstrom build & >= "0.2.0" & < "0.7.0"
  4. digestif build & = "0.5"
  5. nocrypto build
  6. stdint build
  7. jbuilder >= "1.0+beta7"
  8. ocaml >= "4.03.0" & < "4.06.0"

Dev Dependencies


Used by





Innovation. Community. Security.