Numpy 将位打包成 32 位小端值

Question

Numpy provides packbits function to convert from values to individual bits. With bitorder='little' I can read them in C as uint8_t values without issues. However, I would like to read them as uint32_t values. This means that I have to reverse the order of each 4 bytes. I tried to use

import numpy as np

array = np.array([1,0,1,1,0,1,0,1,0,1,0,0,1,0,1,1,0,0,1,1,0,1,0,1,1,0,0,1,0,1,0,1, 
   1,0,0,1,1,0,1,0,1,1,0,0,1,1,1,0,0,1])
array = np.packbits(array, bitorder='little')
array.dtype = np.uint32
array.byteswap(inplace=True)

print(array)

but have the following error:

Traceback (most recent call last):
  File "sample.py", line 5, in <module>
    array.dtype = np.uint32
ValueError: When changing to a larger dtype, its size must be a divisor of the total size in bytes of the last axis of the array.

I have 50 bits in the input. The first chunk of 32 bits written in the little-endian format (earliest input bit is the least significant bit) are 0b10101001101011001101001010101101 = 2846675629 , the second is 0b100111001101011001 = 160601 . So the expected output is

[2846675629 160601]

Answer 1

You can't use array.dtype = np.uint32 as you did, because numpy arrays have to be consecutive in memory.

Instead, you can create a new array of the new type.

import numpy as np

array = np.array([1,0,1,1,0,1,0,1,0,1,0,0,1,0,1,1,0,0,1,1,0,1,0,1,1,0,0,1,0,1,0,1,1,0,0,1,1,0,1,0,1,1,0,0,1,1,1,0,0,1])
array = np.packbits(array, bitorder='little')
array = np.array(array, dtype=np.uint32)
array.byteswap(inplace=True)

print(array)

Answer 2

My first answer fixes the exception.

This answer, relies on this and this

• Pad the array from the right to the nearest power of 2
• Reshape to have some arrays, each array of size 32
• Pack bits PER ARRAY and only then view as unit32 .

import numpy as np
import math


# https://stackoverflow.com/questions/49791312/numpy-packbits-pack-to-uint16-array
# https://stackoverflow.com/questions/36534035/pad-0s-of-numpy-array-to-nearest-power-of-two/36534077


def next_power_of_2(number):
    # Returns next power of two following 'number'
    return 2**math.ceil(math.log(number, 2))


a = np.array([
    1, 0, 1, 1, 0, 1, 0, 1, 0, 1, 0, 0, 1, 0, 1, 1, 0, 0, 1, 1, 0, 1, 0, 1, 1, 0, 0, 1, 0, 1, 0, 1,
    1, 0, 0, 1, 1, 0, 1, 0, 1, 1, 0, 0, 1, 1, 1, 0, 0, 1
])

# a = np.array([
#     0 for _ in range(31)
# ] + [1])
padding_size = next_power_of_2(len(a)) - len(a)
b = np.concatenate([a, np.zeros(padding_size)])
c = b.reshape((-1, 32)).astype(np.uint8)
d = np.packbits(c, bitorder='little').view(np.uint32)

print(d)

output:

[2846675629 160601]