在循环中恢复循环的Python方法是什么

Question

I have an python 2.7 application that I am occasionally interrupting.

I am processing a very large data file. To deal with memory constraints, I have divided up the data file into grids that are identified by an x and y components. Each grid is processed independently.

It takes a very long time to process so occasionally, I need to stop the processing at a certain point. Ideally, I would like to update the y_start and x_start and resume the application at the place I left off (without processing a grid that had already been processed).

The main action occurs within a nested for-loop:

x_start=0.0
x_step=0.05
x_size=10.0
y_start=0.0
y_step=0.05
y_size=10.0

x_ranges = zip(np.arange(x_start,x_size,x_step), np.arange(x_step+x_start,x_size+x_step,x_step))

y_ranges = zip(np.arange(0.0,y_size,y_step), np.arange(y_step,y_size+y_step,y_step))


for x_min,x_max in x_ranges:
    for y_min,y_max in y_ranges:

        doAction()

In the code above, I have the x_start handled. y_start should only be used for when x_min = x_start. For all other values of x, it should start at 0.0.

Here's my proposed solution. Is there a better, more pythonic way:

y_ranges_resume = zip(np.arange(y_start,y_size,y_step),np.arange(y_start+y_step,y_size+y_step,y_step)

for x_min,x_max in x_ranges:
    if x_min == x_start:     
        for y_min,y_max in y_ranges_resume:
            doAction()
    else:
        for y_min,y_max in y_ranges:
            doAction()

Answer 1

I'm not sure about a more pythonic way, but you could rewrite it like this (in any language really):

y_ranges_resume = zip(np.arange(y_start,y_size,y_step),np.arange(y_start+y_step,y_size+y_step,y_step)

for x_min,x_max in x_ranges:
    y_ranges_used = y_ranges
    if x_min == x_start:     
        y_ranges_used = y_ranges_resume

    for y_min,y_max in y_ranges_used:
        doAction()

At least then the inner loop is only written once.

Alternatively you could use the ternary, but I error on the side of easier to read rather than smaller code. But for completeness sake you could write the same thing this way:

y_ranges_resume = zip(np.arange(y_start,y_size,y_step),np.arange(y_start+y_step,y_size+y_step,y_step)

for x_min,x_max in x_ranges:
    for y_min, y_max in y_ranges_resume if x_min == x_start else y_ranges:
        doAction()

Answer 2

You could try list comprehension:

[[doAction() for y_min,y_max in y_ranges_resume] if x_min == x_start else [doAction() for y_min,y_max in y_ranges] for x_min,xmax in x_ranges]

Or you could use a lambda function to simplify the condition

correct_y_range = lambda x : y_ranges_resume if x == x_start else y_ranges
[[doAction() for y_min,y_max in correct_y_range(x_min)] for x_min,x_max in x_ranges]

Another option with the lambda function (for better readability)

correct_y_range = lambda x : y_ranges_resume if x == x_start else y_ranges
for x_min,x_max in x_ranges:
    for y_min, y_max in correct_y_range(x_min):
        doAction()