python - 多处理 [concurrent.futures.ProcessPoolExecutor] 未正确调用函数

Question

各位程序员，你好，我是多处理的新手。我看过这个关于多处理的指南：https ://www.youtube.com/watch?v=fKl2JW_qrso&t=2367s并试图让代码工作。基本上，它的作用是读取一堆图像并将它们保存到processed文件夹中。

zip (130MB) 链接：https ://ufile.io/kt420x7q

整个代码：

import time
import concurrent.futures
from pathlib import Path
from PIL import Image, ImageFilter

parent_path = Path(__file__).parent
output_path = parent_path.joinpath('processed')

def process_image(img_name):
    img = Image.open(img_name)
    img = img.filter(ImageFilter.GaussianBlur(15))
    img.thumbnail(size)
    img.save(f'{output_path.joinpath(img_name)}')
    print(f'{img_name} was processed...')

def main():
    img_names = [
        'photo-1532009324734-20a7a5813719.jpg',
        'photo-1524429656589-6633a470097c.jpg',
        'photo-1530224264768-7ff8c1789d79.jpg',
        'photo-1541698444083-023c97d3f4b6.jpg',
        'photo-1522364723953-452d3431c267.jpg',
        'photo-1513938709626-033611b8cc03.jpg',
        'photo-1493976040374-85c8e12f0c0e.jpg',
        'photo-1504198453319-5ce911bafcde.jpg',
        'photo-1530122037265-a5f1f91d3b99.jpg',
        'photo-1516972810927-80185027ca84.jpg',
        'photo-1550439062-609e1531270e.jpg']

    t1 = time.perf_counter()
    size = (1200, 1200)

##     # Normal Loop Function
##    for img_name in img_names:
##        img = Image.open(img_name)
##        img = img.filter(ImageFilter.GaussianBlur(15))
##        img.thumbnail(size)
##        img.save(f'{output_path.joinpath(img_name)}')
##        print(f'{img_name} was processed...')

    # Multiprocessing Function
    with concurrent.futures.ProcessPoolExecutor() as executor:
        executor.map(process_image, img_names)

    t2 = time.perf_counter()
    print(f'Finished in {t2-t1} seconds')

if __name__ == '__main__':
    main()

##input()

如您所见，有 aNormal Loop Function和 a Multiprocessing Function。MP-function应该做与 完全相同的事情，但Normal Loop Function它似乎无法process_image正确执行功能。

当我运行时：

 # Normal Loop Function
for img_name in img_names:
    img = Image.open(img_name)
    img = img.filter(ImageFilter.GaussianBlur(15))
    img.thumbnail(size)
    img.save(f'{output_path.joinpath(img_name)}')
    print(f'{img_name} was processed...')

它按预期工作，打印进度并将图像保存到processed文件夹：

photo-1532009324734-20a7a5813719.jpg was processed...
photo-1524429656589-6633a470097c.jpg was processed...
photo-1530224264768-7ff8c1789d79.jpg was processed...
photo-1541698444083-023c97d3f4b6.jpg was processed...
photo-1522364723953-452d3431c267.jpg was processed...
photo-1513938709626-033611b8cc03.jpg was processed...
photo-1493976040374-85c8e12f0c0e.jpg was processed...
photo-1504198453319-5ce911bafcde.jpg was processed...
photo-1530122037265-a5f1f91d3b99.jpg was processed...
photo-1516972810927-80185027ca84.jpg was processed...
photo-1550439062-609e1531270e.jpg was processed...
Finished in 11.8887978 seconds

但是当我尝试运行时MP-Function：

# Multiprocessing Function
with concurrent.futures.ProcessPoolExecutor() as executor:
    executor.map(process_image, img_names)

输出是：

Finished in 3.8939473 seconds

在这种情况下，根本不会打印进度，也不会将图像保存到processed文件夹中。此输出来自在 IDLE 中执行代码。由于 IDLE 似乎存在多处理问题，我还尝试通过从 Windows 终端运行脚本来执行该脚本。在这种情况下，没有输出，进程会无限期地运行。

标签： pythonmultithreadingconcurrencymultiprocessingconcurrent.futures