Given the following code (that doesn't work):
while True:
# Snip: print out current state
while True:
ok = get_input("Is this ok? (y/n)")
if ok.lower() == "y": break 2 # This doesn't work :(
if ok.lower() == "n": break
# Do more processing with menus and stuff
Is there a way to make this work? Or do I have do one check to break out of the input loop, then another, more limited, check in the outside loop to break out all together if the user is satisfied?
goto
if you're nestled deep in a lot of loops
My first instinct would be to refactor the nested loop into a function and use return
to break out.
Here's another approach that is short. The disadvantage is that you can only break the outer loop, but sometimes it's exactly what you want.
for a in xrange(10):
for b in xrange(20):
if something(a, b):
# Break the inner loop...
break
else:
# Continue if the inner loop wasn't broken.
continue
# Inner loop was broken, break the outer.
break
This uses the for / else construct explained at: Why does python use 'else' after for and while loops?
Key insight: It only seems as if the outer loop always breaks. But if the inner loop doesn't break, the outer loop won't either.
The continue
statement is the magic here. It's in the for-else clause. By definition that happens if there's no inner break. In that situation continue
neatly circumvents the outer break.
something(a, b)
depends on a
too. The outer loop may run as long as something(a, b)
is not True
.
PEP 3136 proposes labeled break/continue. Guido rejected it because "code so complicated to require this feature is very rare". The PEP does mention some workarounds, though (such as the exception technique), while Guido feels refactoring to use return will be simpler in most cases.
return
is usually the way to go, I've seen quite a few cases where a simple concise ‘break 2
’ statement would just make so much sense. Also, refactor/return
doesn't work the same for continue
. In these cases, numeric break and continue would be easier to follow and less cluttered than refactoring to a tiny function, raising exceptions, or convoluted logic involving setting a flag to break at each nest level. It's a shame Guido rejected it.
break; break
would be nice.
break
through all of the loops. Stupid.
First, ordinary logic is helpful.
If, for some reason, the terminating conditions can't be worked out, exceptions are a fall-back plan.
class GetOutOfLoop( Exception ):
pass
try:
done= False
while not done:
isok= False
while not (done or isok):
ok = get_input("Is this ok? (y/n)")
if ok in ("y", "Y") or ok in ("n", "N") :
done= True # probably better
raise GetOutOfLoop
# other stuff
except GetOutOfLoop:
pass
For this specific example, an exception may not be necessary.
On other other hand, we often have "Y", "N" and "Q" options in character-mode applications. For the "Q" option, we want an immediate exit. That's more exceptional.
done
). (2) raising an exception. Merging them together into a single solution just makes it look complicated. For future readers: EITHER use all the lines involving done
, OR define GetOutOfLoop(Exception)
and raise/except that.
Exception
constructor (e.g. raise Exception('bla bla bla')
) are valid in both Python 2 and Python 3. The former is preferable in this case because we don't want our except
block to catch all exceptions, but only the special exception we're using to exit the loop. If we do things the way you suggest, and then a bug in our code causes an unexpected exception to be raised, it'll be wrongly treated the same as deliberately exiting the loop.
I tend to agree that refactoring into a function is usually the best approach for this sort of situation, but for when you really need to break out of nested loops, here's an interesting variant of the exception-raising approach that @S.Lott described. It uses Python's with
statement to make the exception raising look a bit nicer. Define a new context manager (you only have to do this once) with:
from contextlib import contextmanager
@contextmanager
def nested_break():
class NestedBreakException(Exception):
pass
try:
yield NestedBreakException
except NestedBreakException:
pass
Now you can use this context manager as follows:
with nested_break() as mylabel:
while True:
print "current state"
while True:
ok = raw_input("Is this ok? (y/n)")
if ok == "y" or ok == "Y": raise mylabel
if ok == "n" or ok == "N": break
print "more processing"
Advantages: (1) it's slightly cleaner (no explicit try-except block), and (2) you get a custom-built Exception
subclass for each use of nested_break
; no need to declare your own Exception
subclass each time.
Introduce a new variable that you'll use as a 'loop breaker'. First assign something to it(False,0, etc.), and then, inside the outer loop, before you break from it, change the value to something else(True,1,...). Once the loop exits make the 'parent' loop check for that value. Let me demonstrate:
breaker = False #our mighty loop exiter!
while True:
while True:
if conditionMet:
#insert code here...
breaker = True
break
if breaker: # the interesting part!
break # <--- !
If you have an infinite loop, this is the only way out; for other loops execution is really a lot faster. This also works if you have many nested loops. You can exit all, or just a few. Endless possibilities! Hope this helped!
First, you may also consider making the process of getting and validating the input a function; within that function, you can just return the value if its correct, and keep spinning in the while loop if not. This essentially obviates the problem you solved, and can usually be applied in the more general case (breaking out of multiple loops). If you absolutely must keep this structure in your code, and really don't want to deal with bookkeeping booleans...
You may also use goto in the following way (using an April Fools module from here):
#import the stuff
from goto import goto, label
while True:
#snip: print out current state
while True:
ok = get_input("Is this ok? (y/n)")
if ok == "y" or ok == "Y": goto .breakall
if ok == "n" or ok == "N": break
#do more processing with menus and stuff
label .breakall
I know, I know, "thou shalt not use goto" and all that, but it works well in strange cases like this.
To break out of multiple nested loops, without refactoring into a function, make use of a "simulated goto statement" with the built-in StopIteration exception:
try:
for outer in range(100):
for inner in range(100):
if break_early():
raise StopIteration
except StopIteration: pass
See this discussion on the use of goto statements for breaking out of nested loops.
keeplooping = True
while keeplooping:
# Do stuff
while keeplooping:
# Do some other stuff
if finisheddoingstuff():
keeplooping = False
or something like that.
You could set a variable in the inner loop, and check it in the outer loop immediately after the inner loop exits, breaking if appropriate. I kind of like the GOTO method, provided you don't mind using an April Fool's joke module - it’s not Pythonic, but it does make sense.
This isn't the prettiest way to do it, but in my opinion, it's the best way.
def loop():
while True:
#snip: print out current state
while True:
ok = get_input("Is this ok? (y/n)")
if ok == "y" or ok == "Y": return
if ok == "n" or ok == "N": break
#do more processing with menus and stuff
I'm pretty sure you could work out something using recursion here as well, but I don't know if that's a good option for you.
Keep looping if two conditions are true.
I think this is a more Pythonic way:
dejaVu = True
while dejaVu:
while True:
ok = raw_input("Is this ok? (y/n)")
if ok == "y" or ok == "Y" or ok == "n" or ok == "N":
dejaVu = False
break
while dejaVu:
? You set it to true anyway.
True
conditions to skip two loops, but just one is enough.
if not dejaVu: break
at the bottom and thus exit the main loop? I think the solution is closest to what was asked. +1
There is no way to do this from a language level. Some languages have a goto others have a break that takes an argument, python does not. The best options are: Set a flag which is checked by the outer loop, or set the outer loops condition. Put the loop in a function and use return to break out of all the loops at once. Reformulate your logic.
Credit goes to Vivek Nagarajan, Programmer since 1987
Using Function
def doMywork(data):
for i in data:
for e in i:
return
Using flag
is_break = False
for i in data:
if is_break:
break # outer loop break
for e in i:
is_break = True
break # inner loop break
Factor your loop logic into an iterator that yields the loop variables and returns when done -- here is a simple one that lays out images in rows/columns until we're out of images or out of places to put them:
def it(rows, cols, images):
i = 0
for r in xrange(rows):
for c in xrange(cols):
if i >= len(images):
return
yield r, c, images[i]
i += 1
for r, c, image in it(rows=4, cols=4, images=['a.jpg', 'b.jpg', 'c.jpg']):
... do something with r, c, image ...
This has the advantage of splitting up the complicated loop logic and the processing...
There is a hidden trick in the Python while ... else
structure which can be used to simulate the double break without much code changes/additions. In essence if the while
condition is false, the else
block is triggered. Neither exceptions, continue
or break
trigger the else
block. For more information see answers to "Else clause on Python while statement", or Python doc on while (v2.7).
while True:
#snip: print out current state
ok = ""
while ok != "y" and ok != "n":
ok = get_input("Is this ok? (y/n)")
if ok == "n" or ok == "N":
break # Breaks out of inner loop, skipping else
else:
break # Breaks out of outer loop
#do more processing with menus and stuff
The only downside is that you need to move the double breaking condition into the while
condition (or add a flag variable). Variations of this exists also for the for
loop, where the else
block is triggered after loop completion.
How to break out of multiple loops in Python?
and the answer should have been "It does not work, try something else". I know it fixes the exact given example of the OP, but does not answer their question.
An easy way to turn multiple loops into a single, breakable loop is to use numpy.ndindex
for i in range(n):
for j in range(n):
val = x[i, j]
break # still inside the outer loop!
for i, j in np.ndindex(n, n):
val = x[i, j]
break # you left the only loop there was!
You do have to index into your objects, as opposed to being able to iterate through the values explicitly, but at least in simple cases it seems to be approximately 2-20 times simpler than most of the answers suggested.
In this case, as pointed out by others as well, functional decomposition is the way to go. Code in Python 3:
def user_confirms():
while True:
answer = input("Is this OK? (y/n) ").strip().lower()
if answer in "yn":
return answer == "y"
def main():
while True:
# do stuff
if user_confirms():
break
Another way of reducing your iteration to a single-level loop would be via the use of generators as also specified in the python reference
for i, j in ((i, j) for i in A for j in B):
print(i , j)
if (some_condition):
break
You could scale it up to any number of levels for the loop
The downside is that you can no longer break only a single level. It's all or nothing.
Another downside is that it doesn't work with a while loop. I originally wanted to post this answer on Python - `break` out of all loops but unfortunately that's closed as a duplicate of this one
I'd like to remind you that functions in Python can be created right in the middle of the code and can access the surrounding variables transparently for reading and with nonlocal
or global
declaration for writing.
So you can use a function as a "breakable control structure", defining a place you want to return to:
def is_prime(number):
foo = bar = number
def return_here():
nonlocal foo, bar
init_bar = bar
while foo > 0:
bar = init_bar
while bar >= foo:
if foo*bar == number:
return
bar -= 1
foo -= 1
return_here()
if foo == 1:
print(number, 'is prime')
else:
print(number, '=', bar, '*', foo)
>>> is_prime(67)
67 is prime
>>> is_prime(117)
117 = 13 * 9
>>> is_prime(16)
16 = 4 * 4
By using a function:
def myloop():
for i in range(1,6,1): # 1st loop
print('i:',i)
for j in range(1,11,2): # 2nd loop
print(' i, j:' ,i, j)
for k in range(1,21,4): # 3rd loop
print(' i,j,k:', i,j,k)
if i%3==0 and j%3==0 and k%3==0:
return # getting out of all loops
myloop()
Try running the above codes by commenting out the return
as well.
Without using any function:
done = False
for i in range(1,6,1): # 1st loop
print('i:', i)
for j in range(1,11,2): # 2nd loop
print(' i, j:' ,i, j)
for k in range(1,21,4): # 3rd loop
print(' i,j,k:', i,j,k)
if i%3==0 and j%3==0 and k%3==0:
done = True
break # breaking from 3rd loop
if done: break # breaking from 2nd loop
if done: break # breaking from 1st loop
Now, run the above codes as is first and then try running by commenting out each line containing break
one at a time from the bottom.
Try using an infinite generator.
from itertools import repeat
inputs = (get_input("Is this ok? (y/n)") for _ in repeat(None))
response = (i.lower()=="y" for i in inputs if i.lower() in ("y", "n"))
while True:
#snip: print out current state
if next(response):
break
#do more processing with menus and stuff
# this version uses a level counter to choose how far to break out
break_levels = 0
while True:
# snip: print out current state
while True:
ok = get_input("Is this ok? (y/n)")
if ok == "y" or ok == "Y":
break_levels = 1 # how far nested, excluding this break
break
if ok == "n" or ok == "N":
break # normal break
if break_levels:
break_levels -= 1
break # pop another level
if break_levels:
break_levels -= 1
break
# ...and so on
# this version breaks up to a certain label
break_label = None
while True:
# snip: print out current state
while True:
ok = get_input("Is this ok? (y/n)")
if ok == "y" or ok == "Y":
break_label = "outer" # specify label to break to
break
if ok == "n" or ok == "N":
break
if break_label:
if break_label != "inner":
break # propagate up
break_label = None # we have arrived!
if break_label:
if break_label != "outer":
break # propagate up
break_label = None # we have arrived!
#do more processing with menus and stuff
Here's an implementation that seems to work:
break_ = False
for i in range(10):
if break_:
break
for j in range(10):
if j == 3:
break_ = True
break
else:
print(i, j)
The only draw back is that you have to define break_
before the loops.
What I would personally do is use a boolean that toggles when I am ready to break out the outer loop. For example
while True:
#snip: print out current state
quit = False
while True:
ok = input("Is this ok? (y/n)")
if ok.lower() == "y":
quit = True
break # this should work now :-)
if ok.lower() == "n":
quit = True
break # This should work too :-)
if quit:
break
#do more processing with menus and stuff
probably little trick like below will do if not prefer to refactorial into function
added 1 break_level variable to control the while loop condition
break_level = 0
# while break_level < 3: # if we have another level of nested loop here
while break_level < 2:
#snip: print out current state
while break_level < 1:
ok = get_input("Is this ok? (y/n)")
if ok == "y" or ok == "Y": break_level = 2 # break 2 level
if ok == "n" or ok == "N": break_level = 1 # break 1 level
You can define a variable( for example break_statement ), then change it to a different value when two-break condition occurs and use it in if statement to break from second loop also.
while True:
break_statement=0
while True:
ok = raw_input("Is this ok? (y/n)")
if ok == "n" or ok == "N":
break
if ok == "y" or ok == "Y":
break_statement=1
break
if break_statement==1:
break
My reason for coming here is that i had an outer loop and an inner loop like so:
for x in array:
for y in dont_use_these_values:
if x.value==y:
array.remove(x) # fixed, was array.pop(x) in my original answer
continue
do some other stuff with x
As you can see, it won't actually go to the next x, but will go to the next y instead.
what i found to solve this simply was to run through the array twice instead:
for x in array:
for y in dont_use_these_values:
if x.value==y:
array.remove(x) # fixed, was array.pop(x) in my original answer
continue
for x in array:
do some other stuff with x
I know this was a specific case of OP's question, but I am posting it in the hope that it will help someone think about their problem differently while keeping things simple.
break
instead of continue
would do what you want, wouldn't it? :-)
Solutions in two ways
With an example: Are these two matrices equal/same? matrix1 and matrix2 are the same size, n, two-dimensional matrices.
First solution, without a function
same_matrices = True
inner_loop_broken_once = False
n = len(matrix1)
for i in range(n):
for j in range(n):
if matrix1[i][j] != matrix2[i][j]:
same_matrices = False
inner_loop_broken_once = True
break
if inner_loop_broken_once:
break
Second solution, with a function
This is the final solution for my case.
def are_two_matrices_the_same (matrix1, matrix2):
n = len(matrix1)
for i in range(n):
for j in range(n):
if matrix1[i][j] != matrix2[i][j]:
return False
return True
Trying to minimal changes to the OP's question, I just added a flag before breaking the 1st for loop and check that flag on the outer loop to see if we need to brake once again.
break_2 = False
while True:
# Snip: print out current state
if break_2: break
while True:
ok = get_input("Is this ok? (y/n)")
if ok.lower() == "y": break_2 = True
if break_2: break
if ok.lower() == "n": break
# Do more processing with menus and stuff
I came across this recently and, wanting to avoid a duplicate return statement, which can conceal logical errors, looked at @yak's idea. This works well within nested for loops but is not very elegant. An alternative is to check for the condition before the next loop:
b = None
for a in range(10):
if something(a, b): # should never = True if b is None
break
for b in range(20):
pass
This might not work everywhere but is adaptable and, if required, has the advantage of allowing the condition to be duplicated rather than a potential result.
Success story sharing
return
is the right approach. And the reasoning is that, according to the Zen of Python, "flat is better than nested". We have three levels of nesting here and if that starts to get in the way, it is time to reduce the nesting or at least extract the whole nesting into a function of its own.