Why is list element replacement slower than string element replacement in python? -

i trying replace set of elements in data-structure other value. seems doing such kind of replacement faster in strings in lists in case of python (as revealed benchmarking tests below). can please explain why.

note: these tests performed on using python 2.7.

def string_replace_test(s, chars):     """replaces set of chars 0"""     new = s     c in chars:         new = new.replace(c, '0')     return new  def list_replace_test(s, chars):     """replaces set of chars 0"""     in xrange(len(s)):         if s[a] in chars:             s[a] = '0'  if __name__ == '__main__':     import timeit     s = """         lorem ipsum dolor sit amet, consectetur adipiscing elit. donec         etfringilla purus. pellentesque bibendum urna @ neque consectetur         @ tincidunt nulla luctus. pellentesque augue lacus, interdum id         lectus vitae, laoreet suscipit arcu.         """     s2 = list(s)     chars = ['a', 'e', 'i', 'o', 'u']     print(timeit.timeit("string_replace_test(s, chars)", setup="from __main__ import string_replace_test, s, chars"))     print(timeit.timeit("list_replace_test(s2, chars)", setup="from __main__ import list_replace_test, s2, chars")) 

output:

5.09572291374 49.3243050575 

using range():

5.01253795624 53.2320859432 

since there no list.replace() function, built own, picked slow method.

use list comprehension instead:

def list_replace_test(s, chars):     """replaces set of chars 0"""     return [a if not in chars else '0' in s] 

this still slower string replacement because cannot avoid python loop here.

using set chars helps:

chars = set(chars) 

but fastest way replace individual characters in text different technique altogether. use str.translate() that:

from string import maketrans  map = maketrans('aeiou', '0' * 5) def str_translate(s, map):     return s.translate(map) 

with these changes timings become:

>>> timeit.timeit("list_replace_test(s2, chars)", setup="from __main__ import list_replace_test, s2, chars") 28.60542392730713 >>> timeit.timeit("string_replace_test(s, chars)", setup="from __main__ import string_replace_test, s, chars") 4.002871990203857 >>> timeit.timeit("str_translate(s, map)", setup="from __main__ import str_translate, s, map") 0.7250571250915527 

blowing looped str.replace() calls right out of water too.