[Feature Request] Guard against unnecessary int call on functions that already return integer #14292
Closed as not planned
Description
While cleaning up our code base, I noticed the following anti-patterns (?) which seem pretty well-spread:
num = 0.5
# `round` with `ndigits=None` (default value) already gives int
int(round(num))
# I believe `int(round(num, 0))` gives the same result as `round(num)`
int(round(num, 0))
# The same applies to `len`
int(len([1, 2, 3])
# `sourcery` seems to have a rule to suggest against the following
# WARNING: `//` is the floor division, for negative results their behaviors are different
int(non_negative_number / 2) # -> non_negative_number // 2
# Some `math` functions also give int only (perhaps there are more that I'm missing):
- math.ceil
- math.comb
- math.floor
- math.factorial
- math.gcd
- math.lcm
int(math.ceil(num))Also calling int on round(num, ndigits) (with ndigits not as None) in general might be a bad idea as results may very likely be suprising:
round(2.994, 2) # -> 2.99
round(2.995, 2) # -> 3.0
int(round(2.994, 2)) # -> 2
int(round(2.995, 2)) # -> 3
round(2.994) # -> 3
round(2.995) # -> 3On top of the unnecessary verbosity, this also comes with performance penalty, trying to work on 10 floats for 1_000_000 repetitions:
Time taken for round(): 0.805204 seconds
Time taken for int(round()): 0.946426 seconds
Time taken for math.ceil(): 0.429992 seconds
Time taken for int(math.ceil()): 0.524181 seconds
Test script (by GPT)
import timeit
import math
def round_function(value):
return round(value)
def int_round_function(value):
return int(round(value))
def int_ceil_function(value):
return int(math.ceil(value))
def ceil_function(value):
return math.ceil(value)
# Set up the test data
test_values = [0.1, 1.5, 2.3, 3.7, 4.8, 5.9999, -1.5, -2.3, -3.7, -4.8]
# Number of repetitions for the benchmarking
repetitions = 1_000_000
# Benchmarking round()
time_round = timeit.timeit(
stmt='for v in test_values: round_function(v)',
setup='from __main__ import round_function, test_values',
number=repetitions
)
# Benchmarking int(round())
time_int_round = timeit.timeit(
stmt='for v in test_values: int_round_function(v)',
setup='from __main__ import int_round_function, test_values',
number=repetitions
)
# Benchmarking int(math.ceil())
time_int_ceil = timeit.timeit(
stmt='for v in test_values: int_ceil_function(v)',
setup='from __main__ import int_ceil_function, test_values',
number=repetitions
)
# Benchmarking math.ceil()
time_ceil = timeit.timeit(
stmt='for v in test_values: ceil_function(v)',
setup='from __main__ import ceil_function, test_values',
number=repetitions
)
print(f'Time taken for round(): {time_round:.6f} seconds')
print(f'Time taken for int(round()): {time_int_round:.6f} seconds')
print(f'Time taken for math.ceil(): {time_ceil:.6f} seconds')
print(f'Time taken for int(math.ceil()): {time_int_ceil:.6f} seconds')