trim_extra_whitespace_times

trim_space.py

import re
import string

from time import time


def main():
    # Rust version run with `--release` (for reference), using `only_one_string` from here:
    #   https://stackoverflow.com/a/71864244/10237506
    #   trim_whitespace: 30ms
    
    # Python results (tested on Python 3.10.0 on Mac OS X, Intel processor)
    #   trim_whitespace_v1: 85ms
    #   trim_whitespace_v2: 413ms
    #   trim_whitespace_v3: 87ms

    funcs = [trim_whitespace_v1, trim_whitespace_v2, trim_whitespace_v3]
    func_names = [f.__name__ for f in funcs]
    n = 100_000  # run for this many iterations

    for name in func_names:
        fn = globals()[name]
        start = time()
        for i in range(n):
            actual = fn("1 | This is a “TEST”  | Testing   123 |  H e l l o    world!  ")
        print(f'{name}:   {(time() - start) * 1_000:.0f}ms')

    for input, expected in [
        ("testing123!", "testing123!"),
        ("testing 123!", "testing 123!"),
        ("testing   123!  ", "testing 123!"),
        (" testing  12         3! ", "testing 12 3!")
    ]:
        for name in func_names:
            fn = globals()[name]
            actual = fn(input)
            assert actual == expected, f'func={name}\n  actual={actual!r}\n  expected={expected!r}'


def trim_whitespace_v1(s: str) -> str:
    # remove leading and trailing spaces
    s = s.strip()
    # replace two spaces with a single space
    while '  ' in s:
        s = s.replace('  ', ' ')
    # return the cleaned string
    return s


def trim_whitespace_v2(s: str) -> str:
    return re.sub(r' +', ' ', s.strip())


def trim_whitespace_v3(s: str) -> str:
    # be warned that this removes "all whitespace characters (space, tab,
    # newline, return, form feed)"
    return ' '.join(s.split())


if __name__ == '__main__':
    main()