anycrc 0.8.0

The fastest Python CRC Library

This is a Cython module with bindings to the crcany library. It supports calculating CRC hashes of arbitary sizes as well as updating a CRC hash over time. Anycrc supports all of the CRCs listed in the RevEng CRC Catalogue.

Installation

pip install anycrc

Usage

Use an existing model:

>>> import anycrc
>>> crc32 = anycrc.Model('CRC32')
>>> crc32.calc(b'Hello World!')
472456355

Read the data in chunks:

>>> crc32.update(b'Hello ')
3928882368
>>> crc32.update(b'World!')
472456355

The update method changes the internally stored CRC value, while calc doesn't. You can use get to retrieve the CRC value stored within the object:

>>> crc32.get()
472456355

To specify the starting CRC value:

>>> crc32.set(3928882368)
>>> crc32.calc('World!')
472456355

To go back to the initial value, use:

>>> crc32.reset()

Specify your own CRC parameters:

>>> # width, poly, init, refin, refout, xorout
>>> my_crc = anycrc.CRC(10, 0b0101010101, 0x3ff, True, False, 0)
>>> my_crc.calc('Hello World!')
35

The CRC will be computed in parallel when the input is large enough. If you want to disable parallelism, use:

>>> anycrc.set_parallel(False)

For a list of pre-built models, check models.py. To get a list of the models at any time, use the following command:

python -m anycrc models

The maximum possible CRC width is 128 bits.

Benchmarks

Calculating the CRC32 for lorem ipsum 1 million times:

Module	Time Elapsed (s)	Speed (GiB/s)	Relative
anycrc	0.19	2.14	1.00
zlib	0.28	1.49	1.44
fastcrc	0.80	0.52	4.12
crcmod-plus	0.91	0.46	4.71

Calculating the CRC32 for the text of lorem ipsum repeated 1 million times in a single pass:

Module	Time Elapsed (s)	Speed (GiB/s)	Relative
anycrc	0.13	3.21	1.00
zlib	0.22	1.93	1.66
fastcrc	0.67	0.62	5.18
crcmod-plus	0.67	0.62	5.18

Tested on a 10th generation Intel i7 processor. Parallel performance will depend on your system.