如何正确序列化 python 位数组？答案

【问题标题】：How to properly serialize a python bitarray?如何正确序列化 python 位数组？
【发布时间】：2016-06-23 21:13:43
【问题描述】：

简单地使用 tobytes 和 frombytes 方法并不总是给出如下所示的等价对象。

这似乎源于源位数组的长度不能被八整除。

from bitarray import bitarray  # pip install bitarray


foo = bitarray('0000000000000000000000000000000000000000000000000000000000'
               '0000000000000000000000000000000000000000000000000000000000'
               '0000000000000000000000000000000000000000000000000010000000'
               '0000000000000000000000000000000000000000000000000000000000'
               '0000000000000000000000000000000000000000000000000000000000'
               '0000000000000000000000000000000000000000000000000000000000'
               '0000000000000000000000000000000000000000000000010000000000'
               '0000000000000000000000000000000000000000000000000000000000'
               '0000000000000000000000000000000000000000000000000000000000'
               '0000000000010000000000000000000000000000000000000000000000'
               '0000000000000000000000000000000000000000000000000000000000'
               '0000000000000000000000000000000000000000000000000000000000'
               '0000000000000000000000000000000000000000000000000000000000'
               '0000000000000000000000000000000000000000000000000000000000'
               '0000000000000000000000000000000000000000000000000000000000'
               '000000000000000000000000000000000000000000000001000000',
               endian="little")


bar = bitarray(endian="little")
bar.frombytes(foo.tobytes())
assert(foo.tobytes() == bar.tobytes())  # passes
assert(foo == bar)  # fails

【问题讨论】：

标签： python serialization bitarray

【解决方案1】：

解决办法是保存长度，修剪反序列化的bitarray。

from bitarray import bitarray  # pip install bitarray
import json
from base64 import b64encode, b64decode


def serialize(ba):
    return json.dumps({
        "endian": ba.endian(),
        "bytes": b64encode(ba.tobytes()),
        "len": len(ba)
    })


def deserialize(data):
    data = json.loads(data)
    ba = bitarray(endian=data["endian"])
    ba.frombytes(b64decode(data["bytes"]))
    return ba[:data["len"]]


foo = bitarray('0000000000000000000000000000000000000000000000000000000000'
               '0000000000000000000000000000000000000000000000000000000000'
               '0000000000000000000000000000000000000000000000000010000000'
               '0000000000000000000000000000000000000000000000000000000000'
               '0000000000000000000000000000000000000000000000000000000000'
               '0000000000000000000000000000000000000000000000000000000000'
               '0000000000000000000000000000000000000000000000010000000000'
               '0000000000000000000000000000000000000000000000000000000000'
               '0000000000000000000000000000000000000000000000000000000000'
               '0000000000010000000000000000000000000000000000000000000000'
               '0000000000000000000000000000000000000000000000000000000000'
               '0000000000000000000000000000000000000000000000000000000000'
               '0000000000000000000000000000000000000000000000000000000000'
               '0000000000000000000000000000000000000000000000000000000000'
               '0000000000000000000000000000000000000000000000000000000000'
               '000000000000000000000000000000000000000000000001000000',
               endian="little")


bar = deserialize(serialize(foo))
assert(foo.tobytes() == bar.tobytes())
assert(foo == bar)

【讨论】：

【解决方案2】：

您可以使用 bitarray python 模块直接保存和加载 bitarray

# Save bitarray
ba = bitarray('11100000')
with open('bitarray.bin', 'wb') as f:
        ba.tofile(f)

# Load bitarray
ba = bitarray()
with open('bitarray.bin', 'rb') as f:
        ba.fromfile(f)

【讨论】：