SimpleZstd

This library provides simple Pythonic interface of zstd algorithm like Python's builtin gzip library, by wrapping the zstandard library.

• Note that it matches the same interface of builtin zstd support in Python 3.14+ standard library.

Further more, it provides builtin session support to take advantage of zstd's powerful training mode without additional user effort, by auto collecting training samples (10 by default) during compression/decompression.

In [1]: c = open("poetry.lock").read().encode("utf-8")

In [2]: len(c)
Out[2]: 73946

In [3]: from simple_zstd import compress, decompress

In [4]: for i in range(10):
   ...:     x = compress(c)
   ...:     print(len(x), len(x) / len(c))
   ...:     assert decompress(x) == c
   ...:
22259 0.3010169583209369
22259 0.3010169583209369
22259 0.3010169583209369
22259 0.3010169583209369
22259 0.3010169583209369
7123 0.09632704946853109
7123 0.09632704946853109
7123 0.09632704946853109
7123 0.09632704946853109
7123 0.09632704946853109

Note that user can also choose to have dedicated session instead of default one.

In [1]: c = open("poetry.lock").read().encode("utf-8")

In [2]: from simple_zstd import Session

In [3]: session = Session()

In [4]: x = session.compress(c)

In [5]: assert session.decompress(x) == c

For simpler usage, you can also just import zstd instead of simple_zstd.

Dictionary is exposed under session object. You need to store it for later decompression by passing when creating session.