str.startswith() と re.match() のベンチマーク

bench1.py

# -*- coding: utf-8 -*-

import sys, re
from benchmarker import Benchmarker

try:
    xrange
except:
    xrange = range


N = 100000
for bm in Benchmarker(width=30, cycle=10, extra=1):

    string = "https://www.python.org/"

    with bm.empty():
        for _ in xrange(N):
            pass

    with bm("re.match()"):
        for _ in xrange(N):
            re.match(r'^https://', string)

    with bm("re.compile().search()"):
        rexp = re.compile(r'^https://')
        for _ in xrange(N):
            rexp.match(string)

    with bm("str.startswith() #1"):
        for _ in xrange(N):
            string.startswith(("http://", "https://"))

    with bm("str.startswith() #2"):
        patterns = ("http://", "https://")
        for _ in xrange(N):
            string.startswith(patterns)


### Result example on CPython 3.4.1:
r"""
## Ranking                         real
str.startswith() #1              0.0414 (100.0%) *************************
str.startswith() #2              0.0447 ( 92.6%) ***********************
re.compile().search()            0.0941 ( 44.0%) ***********
re.match()                       0.2490 ( 16.6%) ****

## Ratio Matrix                    real    [01]    [02]    [03]    [04]
[01] str.startswith() #1         0.0414  100.0%  108.0%  227.2%  601.2%
[02] str.startswith() #2         0.0447   92.6%  100.0%  210.5%  556.9%
[03] re.compile().search()       0.0941   44.0%   47.5%  100.0%  264.6%
[04] re.match()                  0.2490   16.6%   18.0%   37.8%  100.0%
"""