图像生成速度慢

Question

我对 python 比较陌生，我构建了一个 Tower of Hanoi 求解器，它将解决方案打印为图像 (https://en.wikipedia.org/wiki/Tower_of_Hanoi)

它按预期工作，但生成图像非常慢。

代码如下：

#!/usr/bin/env python
# -*- coding: utf-8 -*-

import argparse
from PIL import Image


def hanoi(disks, source, helper, target, steps):
    if disks > 0:
        hanoi(disks - 1, source, target, helper, steps)
        target.append(source.pop())
        steps.append([SOURCE[:], HELPER[:], TARGET[:]])
        hanoi(disks - 1, helper, source, target, steps)


def save_image(name):
    print('\nSaving image {}.png'.format(name))
    data = []
    peg = args.disks * 2
    cells = peg * 3 + 40  # 40 is to put some spaces between pegs and the border
    for step in steps:
        for _ in range(5):  # White space
            data.append([1 for _ in range(cells)])

        src = step[0]
        hlp = step[1]
        trg = step[2]
        size = max(len(src), len(hlp), len(trg))
        for _ in range(size - len(src)):
            src.append(0)
        for _ in range(size - len(hlp)):
            hlp.append(0)
        for _ in range(size - len(trg)):
            trg.append(0)
        src.reverse()
        hlp.reverse()
        trg.reverse()

        for s, h, t in zip(src, hlp, trg):
            blanksrc = peg - 2 * s
            blankhlp = peg - 2 * h
            blanktrg = peg - 2 * t
            row = [1 for _ in range(10)]
            row += [1 for _ in range(blanksrc // 2)]
            row += [0 for _ in range(s * 2)]
            row += [1 for _ in range(blanksrc // 2)]
            row += [1 for _ in range(10)]
            row += [1 for _ in range(blankhlp // 2)]
            row += [0 for _ in range(h * 2)]
            row += [1 for _ in range(blankhlp // 2)]
            row += [1 for _ in range(10)]
            row += [1 for _ in range(blanktrg // 2)]
            row += [0 for _ in range(t * 2)]
            row += [1 for _ in range(blanktrg // 2)]
            row += [1 for _ in range(10)]
            data.append(row)

        for _ in range(5):
            data.append([1 for _ in range(cells)])  # White space
        data.append([0 for _ in range(cells)])  # Black line to separate steps

        da = [bit for row in data for bit in row]
        image = Image.new('1', (cells, len(data)))
        image.putdata(da)
        image.save('{}.png'.format(name))


if __name__ == '__main__':
    parser = argparse.ArgumentParser()
    parser.add_argument('-d', '--disks', type=int, default=4,
                        help='Number of disks, default 4')
    parser.add_argument('-f', '--filename', help='Filename', required=True)
    args = parser.parse_args()

    if not args.disks > 0:
        raise ValueError('There must be at least one disk')

    SOURCE = list(reversed(range(1, args.disks + 1)))
    TARGET = []
    HELPER = []
    steps = [[SOURCE[:], HELPER[:], TARGET[:]]]

    hanoi(args.disks, SOURCE, HELPER, TARGET, steps)

    save_image(args.filename)

如前所述，生成图像需要很长时间。

我怎样才能让它更快，为什么它这么慢？

谢谢

Answer 1

用于生成图像的块在 for 循环中。取消缩进，然后将执行时间除以 3。其余的（在我的旧桌面上是 0.04 秒，不是那么长）是 PIL 导入开销。我认为唯一更快的方法是使用 ppm 图像格式（但它会提供巨大的文件）而不是使用 PIL。无论如何，你还是应该听听@PixelEinstein 的建议，试试codereview

图像生成速度慢

Slow image generation

python

image

pillow