如何摆脱python中的多个嵌套循环

Question

This is my code;

instance_count = 10
for imgs in wire_images:
    for counter in range(0, instance_count, 1):
        for oimgs in images_path:
            applyWireAugmentation(oimgs, imgs, wire_dir, odir, 0, dst_dir, counter, "waug")


def applyWireAugmentation(img, wire_img,wdir, odir, theata,dst_path, counter, index):

    src_im = Image.open(wdir+wire_img).convert("LA")
    dst_im = Image.open(odir+img)
    w,h = dst_im.size
    angle = theata
    size = 200, h

    x = random.randint(0,w)
    y = 0
    im = src_im.convert('RGBA')
    rot = im.rotate( angle, expand=1 ).resize(size)
    dst_im.paste( rot, (x, y), rot )
    dst_im.save(dst_path+"/"+img.replace(".png","")+"-"+index+"-"+str(counter)+".png")

wire_images has all the image files that are in a folder. images_path has all the images files in another folder. I need to take one image from wire_images apply it on oimgs instance_count times (10 in this case). Is there any sleek pythonic way to get rid of these loops or to make it faster?

Answer 1

You could generate all the possible combinations before, but I am not sure if it is better:

import itertools
comb = itertools.product(wire_images, images_path, range(0,instance_count,1))

for imgs, oimgs, counter in comb:
    applyWireAugmentation(oimgs, imgs,wire_dir,odir, 0, dst_dir, counter,  "waug")

Answer 2

itertools.product can be helpful here.

It can transform nested for loops to tuples which you can then use in your final statement.

You code will be like this:

from itertools import product   

for imgs, counter, oimgs in product(wire_images, range(0,instance_count,1), images_path):

    applyWireAugmentation(oimgs, imgs, wire_dir, odir, 0, dst_dir, counter, "waug")

Keep in mind that this for loop will not execute if any of the lists(iterators) are empty and it is a bit slower than nested for loop so you have to sacrifice performance a little bit in favor of readability.

Have a look at this Stackoverflow discussion for more info

Answer 3

As others suggested, one of the best possible way to achieve this is to use a generator which reduces the memory overhead drastically since there is no list to hold in-memory.

from itertools import product

wire_images = list1...  # if possible, try to get them in iterator format as well
images_path = list2...   # if possible, try to get them in iterator format as well
instance_count = 10

for imgs, oimgs, counter in product(wire_images, images_path, range(instance_count)):
    applyWireAugmentation(oimgs, imgs, wire_dir, odir, 0, dst_dir, counter, "waug")

The above approach has trade off of time over space and readability.