Python adding additional code to inline for loops?

Question

The following is totally bogus code. But let's say you needed to do some extra side effecting function calls (for debugging to logs)? How would you put that in?

[ i for i in range(10) ]

Or does one always have to rewrite as a normal for loop?

list=[]
for i in range(10):
   otherStuff()
   list.append(i)

In C, there is a comma operator for such things...

IMHO if function is called for it's side effects, explicit loop should be used. If function is called for it's return value, list comprehension should be used. — Łukasz Rogalski, Mar 31 '16 at 15:27
@Chris Oh, ok. However for the sake of readability I wouldn't use that ever. It doesn't seem worth it. — freakish, Mar 31 '16 at 15:32

score 6 · Answer 1 · edited May 23 '17 at 12:07

6

Plainly, don't use side-effects in list comprehensions. It makes your code incredibly unclear to the next person who has to maintain it, even if you understand it perfectly. List comprehensions are a succinct way of creating a list, not a way to call a function n times.

For further reading, see the question Is it Pythonic to use list comprehensions for just side effects?

In other words, you should use an explicit for loop for that.

edited May 23 '17 at 12:07

Community

1
1

answered Mar 31 '16 at 15:28

Adam Smith

52,157
12
73
112

1

@Chris programmer-time versus ease of use is a common trade off. Best practice is to use an explicit `for` loop. YMMV – Adam Smith Mar 31 '16 at 15:30
@Chris You want to debug list comprehension? Doesn't make much sense. Just print it afterwards. – freakish Mar 31 '16 at 15:34

Robᵩ · Accepted Answer · 2016-03-31T16:09:07.293

3

You need to include a call to your side-effect-having code somewhere in your value expression, but you need to ignore that value.

or is one possible choice for this. Just make sure that your side-effect function returns a "Falsey" value (False, None, 0, etc.), and put your debug call in the left-hand side of the or.

def debug_func(i):
    print i, i**3
    return None
whole_numbers = [ debug_func(i) or i for i in range(10) ]
print whole_numbers

As an alternative, your function could be an identity function, always returning its sole argument:

def debug_func(i):
    print i, i**3
    return i

# Production code:
whole_numbers = [i for i in range(10)]

# Debug code
whole_numbers = [debug_func(i) for i in range(10)]

edited Mar 31 '16 at 16:09

answered Mar 31 '16 at 15:31

Robᵩ

163,533
20
239
308

2

Ps. I agree with others. Don't do this. But you didn't ask **if** you should do it, you asked **how** you could do it. – Robᵩ Mar 31 '16 at 15:32
4

It's important that your side-effect function returns a Falsey value here! That's obvious if you've seen this construct before, but less so otherwise :) – Adam Smith Mar 31 '16 at 15:37
This is purely for debugging a pile of list comprehensions. I promise to use sparingly. ;) – Chris Mar 31 '16 at 15:37
@Chris - another option added to answer. – Robᵩ Mar 31 '16 at 16:10

score 1 · Answer 3 · answered Mar 31 '16 at 15:42

1

Here's one option that doesn't require anything about what your function returns:

[(myfunc(), i)[1] for i in range(10)]

You can also do more than one function at a time:

[(myfunc(), myfunc2(), i)[-1] for i in range(10)]

answered Mar 31 '16 at 15:42

zondo

19,901
8
44
83

Python adding additional code to inline for loops?

3 Answers3