I examined how to delete the substring obtained by using re in python from the original string, so instead of a memo.
The following two things came to my mind
If you feel that you have tested several patterns using timeit, 1 seems to be faster, so I adopted this.
The code used for the test is as follows.
import timeit
import re
test_str = '...Something string...'
pat = re.compile('...Something regular expression...')
match = pat.search(test_str)
def _remove_str1(base_str, target):
    """ match_base based on object data_Disconnect str and reconnect"""
    start = target.start()
    end = target.end()
    return base_str[:start] + base_str[end:]
def _remove_str2(base_str, target):
    """ match_Based on object data
    base_Replace the relevant part of str with an empty string"""
    return base_str.replace(target.group(), '') 
def run1():
    _remove_str1(test_str, match)
def run2():
    _remove_str2(test_str, match)
if __name__ == '__main__':
    t = timeit.Timer('run1()',
            'from %s import run1' % __name__)
    print t.timeit()
    t2 = timeit.Timer('run2()',
            'from %s import run2' % __name__)
    print t2.timeit()