How to compose a list with conditional elements

Question

I program in python for a while and I've found that this language is very programmer friendly, so that maybe there is a technique that I don't know how to compose a list with conditional elements. The simplified example is:

# in pseudo code
add_two = True
my_list = [
  "one",
  "two" if add_two,
  "three",
]

Basically I'm looking for a handy way to create a list that contains some that are added under some specific condition.

Some alternatives that don't look such nice:

add_two = True

# option 1
my_list = []
my_list += ["one"]
my_list += ["two"] if add_two else []
my_list += ["three"]


# option 2
my_list = []
my_list += ["one"]
if add_two: my_list += ["two"]
my_list += ["three"]

Is there something that can simplify it? Cheers!

Answer 1

If you can create a list of bools representing what elements you want to keep from a candidate list, you can do this pretty succinctly. For example:

candidates = ['one', 'two', 'three', 'four', 'five']
include = [True, True, False, True, False]
result = [c for c, i in zip(candidates, include) if i]
print(result)
# ['one', 'two', 'four']

If you can use numpy, this gets even more succinct:

import numpy as np
candidates = np.array(['one', 'two', 'three', 'four', 'five'])
include = [True, True, False, True, False]
print(candidates[include])  # can use boolean indexing directly!
# ['one', 'two', 'four']

Finally, as suggested in a comment, you can use itertools.compress() . Note that this returns an iterator, so you have to unpack it.

from itertools import compress
print([v for v in compress(candidates, include)])
# ['one', 'two', 'four']

Answer 2

In one line you can write:

my_list = ['one'] + (['two'] if add_two else []) + ['three']

Or use a list comprehension:

my_list = [x for x in ('one', 'two' if add_two else '', 'three') if x]

Or the functional way to remove Falsy values:

my_list = list(filter(None, ('one', 'two' if add_two else '', 'three')))

Answer 3

First I'd write a simple predicate function that determines whether or not a value should be included. Let's pretend that in this list of integers you only want to include those numbers >0 .

def is_strictly_positive(x):
    return x > 0

Then you can have your whole list of numbers:

lst = [-3, -2, -1, 0, 1, 2, 3, 4, 5, 6, 7]

And filter them:

filter(is_strictly_positive, lst)  # 1 2 3 4 5 6 7

which creates a filter object -- a generator that produces the values you need once. If you need a whole list, you can either do:

new_lst = list(filter(is_strictly_positive, lst))  # [1, 2, 3, 4, 5, 6, 7]

or, idiomatically, use a list comprehension

new_lst = [x for x in lst if is_strictly_positive(x)]  # [1, 2, 3, 4, 5, 6, 7]

---

You can also use itertools.compress to produce a similar result to filter , but it's a little over-engineered in this simple case.

new_lst_gen = itertools.compress(lst, map(is_strictly_positive, lst))  # 1 2 3 4 5 6 7

Answer 4

This approach uses a None sentinel value for values to remove, then filters them out at the end. If your data contains None already, you can create another sentinel object to use instead.

add_two = True
my_list = [
    "one",
    "two" if add_two else None,
    "three",
]

my_list = [e for e in my_list if e is not None]

print(my_list)
# ['one', 'two', 'three']

Answer 5

The itertools module has many useful functions:

from itertools import compress, dropwhile, takewhile, filterfalse

condition1 = True
condition2 = False
condition3 = True
l = list(range(10))

print(list(compress(['a', 'b', 'c'], [condition1, condition2, condition3])))
# ['a', 'c']

print(list(dropwhile(lambda x: x > 5, l)))
# [5, 6, 7, 8, 9]

print(list(takewhile(lambda x: x < 5, l)))
# [0, 1, 2, 3, 4]

print(list(filterfalse(lambda x: x % 2, l))) # returns elements whose predicates == False
# [0, 2, 4, 6, 8]

Answer 6

Another one line approach, which is more Pythonic in my opinion, would be:

add_two = True    
my_list = ['one'] + ['two'] * add_two + ['three']

In case of multiple conditionned elements, I think other alternatives, such as list of bools, should be used.