What is the most efficient way in Python to convert a string to all lowercase stripping out all non-ascii alpha characters?

Another solution (not that pythonic, but very fast) is to use string.translate – though note that this will not work for unicode. It’s also worth noting that you can speed up Dana’s code by moving the characters into a set (which looks up by hash, rather than performing a linear search each time). Here are the timings I get for various of the solutions given:

import string, re, timeit

# Precomputed values (for str_join_set and translate)

letter_set = frozenset(string.ascii_lowercase + string.ascii_uppercase)
tab = string.maketrans(string.ascii_lowercase + string.ascii_uppercase,
                       string.ascii_lowercase * 2)
deletions="".join(ch for ch in map(chr,range(256)) if ch not in letter_set)

s="A235th@#$&( er Ra{}|?>ndom"

# From unwind's filter approach
def test_filter(s):
    return filter(lambda x: x in string.ascii_lowercase, s.lower())

# using set instead (and contains)
def test_filter_set(s):
    return filter(letter_set.__contains__, s).lower()

# Tomalak's solution
def test_regex(s):
    return re.sub('[^a-z]', '', s.lower())

# Dana's
def test_str_join(s):
    return ''.join(c for c in s.lower() if c in string.ascii_lowercase)

# Modified to use a set.
def test_str_join_set(s):
    return ''.join(c for c in s.lower() if c in letter_set)

# Translate approach.
def test_translate(s):
    return string.translate(s, tab, deletions)


for test in sorted(globals()):
    if test.startswith("test_"):
        assert globals()[test](s)=='atherrandom'
        print "%30s : %s" % (test, timeit.Timer("f(s)", 
              "from __main__ import %s as f, s" % test).timeit(200000))

This gives me:

               test_filter : 2.57138351271
           test_filter_set : 0.981806765698
                test_regex : 3.10069885233
             test_str_join : 2.87172979743
         test_str_join_set : 2.43197956381
            test_translate : 0.335367566218

[Edit] Updated with filter solutions as well. (Note that using set.__contains__ makes a big difference here, as it avoids making an extra function call for the lambda.