Python 将字节数组转换为元组数组的快速方法（不使用 NumPy）

Question

我需要加速以下代码：-

index = 0
WIDTH = 1920
HEIGHT = 1080
for o in xrange(WIDTH * HEIGHT):
    outbuf[o] = (buffer[index], buffer[index+1], buffer[index+2])
    index += 3

将字节数组转换为元组数组，但速度太慢，有没有办法在不使用 numpy 的情况下使其更快。

Answer 1

使用 itertools 中的 grouper recipe 构建 outbuf 在我的机器上几乎快两倍。

这是文档中的 grouper 函数：

def grouper(iterable, n, fillvalue=None):
    "Collect data into fixed-length chunks or blocks"
    # grouper('ABCDEFG', 3, 'x') --> ABC DEF Gxx
    args = [iter(iterable)] * n
    return izip_longest(fillvalue=fillvalue, *args)

此脚本比较问题中函数的计时，使用 grouper 填充 outbuf 并使用 grouper 填充 return outbuf 而不是填充外部列表。最快的方法是从函数return outbuf.

from itertools import izip_longest
import random
import sys
import timeit

WIDTH = 1920
HEIGHT = 1080


buffer_ = bytearray(random.randint(0, 255) for _ in xrange(WIDTH * HEIGHT * 3))

# Function from the question
def extract(buffer_):
    index = 0
    for o in xrange(WIDTH * HEIGHT):
        outbuf[o] = (buffer_[index], buffer_[index+1], buffer_[index+2])
        index += 3


def grouper(iterable, n, fillvalue=None):
    args = [iter(iterable)] * n
    return izip_longest(fillvalue=fillvalue, *args)


# Use grouper with external outbuf
def grouper_extract(buffer_):
    groups = grouper(buffer_, 3)
    for o in xrange(WIDTH * HEIGHT):
        outbuf[o] = next(groups)


# Return a list using grouper
def grouper_list(buffer_):
    return list(grouper(buffer_, 3))


if __name__ == '__main__':
    # Set number of timeit repetitions.
    try:
        number = int(sys.argv[1])
    except IndexError:
        number = 50
    outbuf = [0 for _ in xrange(WIDTH * HEIGHT * 3)]
    print 'OP function'
    print timeit.timeit(setup="from __main__ import extract, outbuf, buffer_",
                        stmt="extract(buffer_)", number=number)
    print
    print 'Fill external outbuf with grouper'
    print timeit.timeit(setup="from __main__ import grouper_extract, outbuf, buffer_",
                        stmt="grouper_extract(buffer_)", number=number)
    print
    print 'Return outbuf using grouper'
    print timeit.timeit(setup="from __main__ import grouper_list, buffer_",
                        stmt="outbuf = grouper_list(buffer_)", number=number)
    print

以下是每种方法重复 50 次的时间安排：

OP function
39.3345730305

Fill external outbuf with grouper
30.0249710083

Return outbuf using grouper
20.357350111

Python 将字节数组转换为元组数组的快速方法（不使用 NumPy）

In Python fast way to convert a bytearray to array of tuples (Without using NumPy)

python

arrays

tuples

python-2.x