tibs 2.0.1

A sleek Python library for binary data

tibs is a Python library for binary data. It's 100% written in Rust and has excellent performance.

Use it for packets, registers, instruction formats, bitsets, compressed data and streams where fields can have many different interpretations and be any number of bits long.

It is used to power the popular bitstring library, which is by the same author. The full documentation is available on Read the Docs.

Install

pip install tibs

Tibs works with Python 3.11 and later. There are pre-built wheels for most common platforms; if there are issues then please let me know.

Overview

The tibs library provides two main classes: Tibs, which is an immutable sequence of bits (similar to how bytes works in Python as a sequence of bytes) and Mutibs, which is a mutable version (similar to bytearray in Python).

They can be used in a few ways, depending on what you need:

1. As a container of bits

Tibs provides an interface very similar to bytes and other Python containers - you can slice it, concatenate, search it etc. in a familiar way, with Mutibs adding on mutating methods.

find · rfind · find_all · replace · count · starts_with · split_at · chunks · + · in

This 'container of bits' mental model might be all that you need, but the library also gives you two broad views of the binary data.

2. As Typed fields

Pull integers, floats, bytes, hex or binary of any bit length straight out of the bits, without hand-rolling shifts and masks. Little-endian ordering and LSB0 field labels are handled elegantly so you don't reshuffle data yourself, and extracted / deposit reach fields that are scattered across a word.

When parsing streams, a Reader can wrap the Tibs to hold a bit position for you, so a parsing loop never has to work out where the next one starts.

from_u · to_f · bin / hex · Dtype · pack / unpack · .le · .lsb0 · field() · extracted / deposit · Reader · f-string formatting

3. As a set of bits

For bitwise algebra, cardinalities and set predicates, with no intermediate object built along the way. Mutibs can also be used as a large mutable bitset.

& | ^ ~ · count_and · count_xor · intersects · is_subset_of · set / unset · all / any

And it's fast — usually significantly faster than similar libraries.

A Taster

Some real code to illustrate.

As a container of bits. Tibs works like bytes, except that the unit is the bit. Mutibs is its mutable counterpart, for patching in place.

>>> from tibs import Tibs
>>> # A 5-bit header, a message, then 3 bits of padding: nothing is byte aligned.
>>> frame = Tibs('0b10110') + b'the cat rarely blinked' + [0, 0, 0]
>>> bytes(frame).find(b'cat')      # using bytes, the message can't be found
-1
>>> pos = frame.find(b'cat')       # but the tibs still knows where it is
>>> pos, frame[pos:pos + 24].bytes
(37, b'cat')

>>> patched = frame.to_mutibs()
>>> patched[pos:pos + 24] = b'squirrel'
>>> patched[5:-3].bytes
b'the squirrel rarely blinked'
>>> len(frame), len(patched)       # 40 bits longer, spliced in at bit 37
(184, 224)

As typed fields. Read and write integers, floats and strings of any bit length, with a view taking care of byte order and bit numbering — the sort of job that gets awkward quickly with plain bytes and masks.

>>> # What's inside a float? A sign bit, an 8-bit exponent and a 23-bit fraction.
>>> x = Tibs.from_f(-118.625, 32)
>>> f"{x:_.8b}"                    # grouped into bytes to make it readable
'11000010_11101101_01000000_00000000'
>>> sign, exponent, fraction = x.split_at([1, 9])
>>> (-1) ** sign.u * 2 ** (exponent.u - 127) * (1 + fraction.u / 2 ** 23)
-118.625

>>> Tibs(b'\x00\x40\xed\xc2').le.f     # the same value, from a little-endian file
-118.625
>>> Tibs.from_u(x.u + 1, 32).f         # the adjacent float32, one bit away
-118.62500762939453

As a set of bits. Bitwise algebra and cardinalities over millions of bits, without building an intermediate object just to count it.

>>> from math import isqrt
>>> from tibs import Mutibs
>>> # A sieve of Eratosthenes over ten million numbers, one bit each.
>>> limit = 10_000_000
>>> sieve = Mutibs.from_ones(limit)
>>> sieve.unset([0, 1])
>>> for p in range(2, isqrt(limit) + 1):
...     if sieve[p]:
...         sieve.unset(range(p * p, limit, p))
...
>>> sieve.count(1)                     # primes below ten million
664579

>>> # Counting twin, cousin and sexy primes: pairs 2, 4 and 6 apart:
>>> [sieve.count_and(sieve >> d) for d in (2, 4, 6)]
[58980, 58622, 117207]

The full documentation covers construction, interpretation, endianness, searching and replacing, indexing, serialization, views, dtypes and much more.

Performance

Tibs is written in Rust with PyO3. The repository contains a dedicated performance regression suite and CI workflow that compare benchmark medians against the base commit.

For local comparisons, tests/performance_comparison.py checks common operations against the bitarray library and the standard Python library. With bitarray installed, run:

python tests/performance_comparison.py

Benchmarks are machine-dependent, but tibs is often almost unreasonably fast.

Examples

The examples are small, but they are meant to look like real binary-data tasks. Each is walked through in the documentation, with the runnable code in examples/. Some examples of the examples:

The rest of the examples cover stream scanning, in-place patching, structured headers, bulk sample packing and scattered register fields.

Project status

Tibs is considered 'stable' and has reached version 2. Documented public behavior will remain compatible across future 2.x releases. It is already used to power the bitstring library and gets several million downloads per month.

There are thousands of unit tests, including Hypothesis tests and performance benchmarks.

For the full API reference, see the documentation.

Credits

The tibs library was created by Scott Griffiths and is released under the MIT License.

The Tibs cat artwork was created by Ada Griffiths and is not covered by the software license. All rights reserved.