This library is inspired by Segment's KSUID implementation: https://github.com/segmentio/ksuid
For the Rust version, please check out https://github.com/svix/rust-ksuid
A ksuid is a K sorted UID. In other words, a KSUID also stores a date component, so that ksuids can be approximately sorted based on the time they were created.
Read more here.
pip install svix-ksuid
from ksuid import Ksuid
ksuid = Ksuid()Ksuids have a 1 second accuracy which is not sufficient for all use-cases. That's why this library exposes a higher accuracy mode which supports accuracy of up to 4ms.
It's fully compatible with normal ksuids, in fact, it outputs valid ksuids. The difference is that it sacrifices one byte of the random payload in favor of this accuracy.
The code too is fully compatible:
pip install svix-ksuid
from ksuid import KsuidMs
ksuid = KsuidMs()Generate a ksuid without passing a specific datetime
In [1]: from ksuid import Ksuid
In [2]: ksuid = Ksuid()
In [3]: f"Base62: {ksuid}"
Out[3]: 'Base62: 1srOrx2ZWZBpBUvZwXKQmoEYga2'
In [4]: f"Bytes: {bytes(ksuid)}"
Out[4]: "Bytes: b'\\r5\\xc43\\xe1\\x93>7\\xf2up\\x87c\\xad\\xc7tZ\\xf5\\xe7\\xf2'"
In [5]: f"Datetime: {ksuid.datetime}"
Out[5]: 'Datetime: 2021-05-21 14:04:03'
In [6]: f"Timestamp: {ksuid.timestamp}"
Out[6]: 'Timestamp: 1621627443'
In [7]: f"Payload: {ksuid.payload}"
Out[7]: "Payload: b'\\xe1\\x93>7\\xf2up\\x87c\\xad\\xc7tZ\\xf5\\xe7\\xf2'"In [1]: datetime = datetime(year=2021, month=5, day=19, hour=1, minute=1, second=1, microsecond=1)
In [2]: datetime
Out[2]: datetime.datetime(2021, 5, 19, 1, 1, 1, 1)
In [3]: ksuid = Ksuid(datetime)
In [4]: ksuid.datetime
Out[4]: datetime.datetime(2021, 5, 19, 1, 1, 1)
In [5]: ksuid.timestamp
Out[5]: 1621407661In [1]: ksuid = Ksuid()
In [2]: ksuid.timestamp
Out[2]: 1621634852
In [3]: f"Base62: {ksuid}"
Out[3]: 'Base62: 1srdszO8Xy2cR6CnARnvxCfRmK4'
In [4]: ksuid_from_base62 = Ksuid.from_base62("1srdszO8Xy2cR6CnARnvxCfRmK4")
In [5]: ksuid_from_base62.timestamp
Out[5]: 1621634852In [1]: ksuid = Ksuid()
In [2]: ksuid_from_bytes = ksuid.from_bytes(bytes(ksuid))
In [3]: f"ksuid: {ksuid}, ksuid_from_bytes: {ksuid_from_bytes}"
Out[3]: 'ksuid: 1sreAHoz6myPhXghsOdVBoec3Vr, ksuid_from_bytes: 1sreAHoz6myPhXghsOdVBoec3Vr'
In [4]: ksuid == ksuid_from_bytes
Out[4]: TrueIn [1]: ksuid_1 = Ksuid()
In [2]: ksuid_2 = Ksuid.from_bytes(bytes(ksuid_1))
In [3]: f"ksuid_1: {ksuid_1}, ksuid_2: {ksuid_2}"
Out[3]: 'ksuid_1: 1sreAHoz6myPhXghsOdVBoec3Vr, ksuid_2: 1sreAHoz6myPhXghsOdVBoec3Vr'
In [4]: ksuid_1 == ksuid_2
Out[4]: True
In [5]: ksuid_1
Out[5]: 1tM9eRSTrHIrrH5SMEW24rtvIOF
In [6]: ksuid_2
Out[6]: 1tM9eRSTrHIrrH5SMEW24rtvIOFIn [1]: ksuid_1 = Ksuid()
In [2]: ksuid_1.timestamp
Out[2]: 1621963256
In [3]: ksuid_2 = Ksuid()
In [4]: ksuid_2.timestamp
Out[4]: 1621963266
In [5]: ksuid_1 < ksuid_2
Out[5]: True
In [6]: ksuid_1 <= ksuid_2
Out[6]: True
In [7]: ksuid_1 >= ksuid_2
Out[7]: False
In [8]: ksuid_1 > ksuid_2
Out[8]: Falseksuid source code is available under an MIT License.