The original title was “Never Write For-Loops Again” but I think it misled people to think that for-loops are bad. This wasn’t my intent. This article provides several alternatives for cases, IMHO, don’t need explicit for-loops, and I think it’s better not writing them, or at least, do a quick mental exercise to think of an alternative.
There are several ways to re-write for-loops in Python. Pause yourself when you have the urge to write a for-loop next time. Ask yourself, “Do I really need a for-loop to express the idea? Or is there a even more expressive way?” This article isn’t trying to be dictating the way you think about writing code. Instead, this article merely provides you a different perspective. Hopefully, you’ll get shocked and learn something new.
Quite Shocking, huh? This is a challenge. I challenge you to avoid writing for-loops in every scenario. Also, I challenge you to find the scenarios that are so freaking hard to write anything else but a for-loop. Please share your findings. I’d like to hear about them.
It’s been a while since I started exploring the amazing language features in Python. At the beginning, it’s just a challenge I gave myself to practice using more language features instead of those I learned from other programming language. And things are just getting more fun! Not only the code become shorter and cleaner, but also code looks more structured and disciplined. I’ll get into those benefits more in this article.
But first, let’s take a step back and see what’s the intuition behind writing a for-loop:
- To go through a sequence to extract out some information
- To generate another sequence out of the current sequence
- This is my second nature to write for-loops because I’m a programmer
Fortunately, there are already great tools that are built into Python to help you accomplish the goals! All you need is to shift your mind and look at the things in a different angle.
What you gain by not writing for-loops everywhere
- Fewer lines of code
- Better code readability
- Leave indentation for managing context only
Let’s see the code skeleton below:
with ...:
for ...:
if ...:
try:
except:
else:
In this example, we are dealing with multiple layers of code. THIS IS HARD TO READ. The problem I found in this code is that it is mixing the administrative logic (the with, try-except) with the business logic (the for, if) by giving them the indentation ubiquitously. If you are disciplined about using indentation only for administrative logic, your core business logic would stand out immediately.
“Flat is better than nested” — The Zen of Python
“I wish the code is flatter,” I hear you.
Tools you can use to avoid using for-loops
1. List Comprehension / Generator Expression
Let’s see a simple example. Basically you want to compile a sequence based on another existing sequence:
result = []
for item in item_list:
new_item = do_something_with(item)
result.append(new_item)
You can use map if you love MapReduce, or, Python has List Comprehension:
result = [do_something_with(item) for item in item_list]
Similarly, if you wish to get a iterator only, you can use Generator Expression with almost the same syntax. (How can you not love the consistency in Python?)
result = (do_something_with(item) for item in item_list)
2. Functions
Thinking in a higher-order, more functional programming way, if you want to map a sequence to another, simply call the map function. (Be my guest to use list comprehension here instead.)
doubled_list = map(lambda x: x * 2, old_list)
If you want to reduce a sequence into a single value, use reduce
from functools import reduce
summation = reduce(lambda x, y: x + y, numbers)
Also, lots of Python’s builtin functions consumes iterables (sequences are all iterable by definition):
>>> a = list(range(10))
>>> a
[0, 1, 2, 3, 4, 5, 6, 7, 8, 9]
>>> all(a)
False
>>> any(a)
True
>>> max(a)
9
>>> min(a)
0
>>> list(filter(bool, a))
[1, 2, 3, 4, 5, 6, 7, 8, 9]
>>> set(a)
{0, 1, 2, 3, 4, 5, 6, 7, 8, 9}
>>> dict(zip(a,a))
{0: 0, 1: 1, 2: 2, 3: 3, 4: 4, 5: 5, 6: 6, 7: 7, 8: 8, 9: 9}
>>> sorted(a, reverse=True)
[9, 8, 7, 6, 5, 4, 3, 2, 1, 0]
>>> str(a)
‘[0, 1, 2, 3, 4, 5, 6, 7, 8, 9]’
>>> sum(a)
45
3. Extract Functions or Generators
The above two methods are great to deal with simpler logic. How about more complex logic? As a programmer, we write functions to abstract out the difficult things. Same idea applies here. If you are writing this:
results = []
for item in item_list:
# setups
# condition
# processing
# calculation
results.append(result)
Apparently you are giving too much responsibility to a single code block. Instead, I propose you do:
def process_item(item):
# setups
# condition
# processing
# calculation
return resultresults = [process_item(item) for item in item_list]
How about nested for-loops?
results = []
for i in range(10):
for j in range(i):
results.append((i, j))
List Comprehension got your back:
results = [(i, j)
for i in range(10)
for j in range(i)]
How about if you have some internal state in the code block to keep?
# finding the max prior to the current item
a = [3, 4, 6, 2, 1, 9, 0, 7, 5, 8]
results = []
current_max = 0
for i in a:
current_max = max(i, current_max)
results.append(current_max)# results = [3, 4, 6, 6, 6, 9, 9, 9, 9, 9]
Let’s extract a generator to achieve this:
def max_generator(numbers):
current_max = 0
for i in numbers:
current_max = max(i, current_max)
yield current_maxa = [3, 4, 6, 2, 1, 9, 0, 7, 5, 8]
results = list(max_generator(a))
“Oh wait, you just used a for-loop in the generator function. That’s cheating!”
Fine, let’s try the following.
4. Don’t write it yourself. itertools got you covered
This module is simply brilliant. I believe this module covers 80% of the cases that you makes you want to write for-loops. For example, the last example can be rewritten to:
from itertools import accumulate
a = [3, 4, 6, 2, 1, 9, 0, 7, 5, 8]
results = list(accumulate(a, max))
I know, I know. This was a terrible example. I was just trying to prove a point — “for-loops could be eliminated in your code.” However, this doesn’t the elimination any better. This example is very convoluted and hard to digest and will make your colleagues hate you for showing off.
Also, if you are iterating on combinatoric sequences, there are product(), permutations(), combinations() to use.
Conclusion
- You don’t need to write for-loops in most scenarios
- You should avoid writing for-loops, so you have better code readability
Action
- Look at your code again. Spot any places that you wrote a for-loop previously by intuition. Think again and see if it make sense to re-write it without using for-loop.
- Share your cases that are hard to code without using for-loops
Sharing Knowledge and programming 