J'ai comparé ces différentes techniques sous Python 3.7.0 (IPython).
TLDR
- le plus rapide (lorsque le symbole de scission
c
est connu): regex pré-compilée.
- le plus rapide (autrement):
s.partition(c)[0]
.
- safe (c'est-à-dire, quand
c
peut ne pas être dans s
): partition, split.
- unsafe: index, regex.
Code
import string, random, re
SYMBOLS = string.ascii_uppercase + string.digits
SIZE = 100
def create_test_set(string_length):
for _ in range(SIZE):
random_string = ''.join(random.choices(SYMBOLS, k=string_length))
yield (random.choice(random_string), random_string)
for string_length in (2**4, 2**8, 2**16, 2**32):
print("\nString length:", string_length)
print(" regex (compiled):", end=" ")
test_set_for_regex = ((re.compile("(.*?)" + c).match, s) for (c, s) in test_set)
%timeit [re_match(s).group() for (re_match, s) in test_set_for_regex]
test_set = list(create_test_set(16))
print(" partition: ", end=" ")
%timeit [s.partition(c)[0] for (c, s) in test_set]
print(" index: ", end=" ")
%timeit [s[:s.index(c)] for (c, s) in test_set]
print(" split (limited): ", end=" ")
%timeit [s.split(c, 1)[0] for (c, s) in test_set]
print(" split: ", end=" ")
%timeit [s.split(c)[0] for (c, s) in test_set]
print(" regex: ", end=" ")
%timeit [re.match("(.*?)" + c, s).group() for (c, s) in test_set]
Résultats
String length: 16
regex (compiled): 156 ns ± 4.41 ns per loop (mean ± std. dev. of 7 runs, 10000000 loops each)
partition: 19.3 µs ± 430 ns per loop (mean ± std. dev. of 7 runs, 100000 loops each)
index: 26.1 µs ± 341 ns per loop (mean ± std. dev. of 7 runs, 10000 loops each)
split (limited): 26.8 µs ± 1.26 µs per loop (mean ± std. dev. of 7 runs, 10000 loops each)
split: 26.3 µs ± 835 ns per loop (mean ± std. dev. of 7 runs, 10000 loops each)
regex: 128 µs ± 4.02 µs per loop (mean ± std. dev. of 7 runs, 10000 loops each)
String length: 256
regex (compiled): 167 ns ± 2.7 ns per loop (mean ± std. dev. of 7 runs, 10000000 loops each)
partition: 20.9 µs ± 694 ns per loop (mean ± std. dev. of 7 runs, 10000 loops each)
index: 28.6 µs ± 2.73 µs per loop (mean ± std. dev. of 7 runs, 10000 loops each)
split (limited): 27.4 µs ± 979 ns per loop (mean ± std. dev. of 7 runs, 10000 loops each)
split: 31.5 µs ± 4.86 µs per loop (mean ± std. dev. of 7 runs, 10000 loops each)
regex: 148 µs ± 7.05 µs per loop (mean ± std. dev. of 7 runs, 1000 loops each)
String length: 65536
regex (compiled): 173 ns ± 3.95 ns per loop (mean ± std. dev. of 7 runs, 10000000 loops each)
partition: 20.9 µs ± 613 ns per loop (mean ± std. dev. of 7 runs, 100000 loops each)
index: 27.7 µs ± 515 ns per loop (mean ± std. dev. of 7 runs, 10000 loops each)
split (limited): 27.2 µs ± 796 ns per loop (mean ± std. dev. of 7 runs, 10000 loops each)
split: 26.5 µs ± 377 ns per loop (mean ± std. dev. of 7 runs, 10000 loops each)
regex: 128 µs ± 1.5 µs per loop (mean ± std. dev. of 7 runs, 10000 loops each)
String length: 4294967296
regex (compiled): 165 ns ± 1.2 ns per loop (mean ± std. dev. of 7 runs, 10000000 loops each)
partition: 19.9 µs ± 144 ns per loop (mean ± std. dev. of 7 runs, 100000 loops each)
index: 27.7 µs ± 571 ns per loop (mean ± std. dev. of 7 runs, 10000 loops each)
split (limited): 26.1 µs ± 472 ns per loop (mean ± std. dev. of 7 runs, 10000 loops each)
split: 28.1 µs ± 1.69 µs per loop (mean ± std. dev. of 7 runs, 10000 loops each)
regex: 137 µs ± 6.53 µs per loop (mean ± std. dev. of 7 runs, 10000 loops each)
s1.partition(':')[0]