How to sort os.walk(path) in alphanumeric order, with duplicates coming after the original file, using Python 3?

In python 3 (specifically 3.10.6), how can you change the way that os.walk(path) sorts the files it finds? Given this list of files:

IMG0001.jpg
IMG0002.jpg
IMG0002(1).jpg
IMG0002(2).jpg
IMG0003.jpg

How would you sort it in that order, with each (n) duplicate file coming after the original file? Currently, os.walk(path) is sorting this list like this:

IMG0001.jpg
IMG0002(1).jpg
IMG0002(2).jpg
IMG0002.jpg
IMG0003.jpg

I suppose the main issue is that the default sort method is giving a higher "sort value" to the ( (and also -) than it is to the . in the extension. If this is correct in what's happening here, how would you modify which special characters come before others?

I've tried to use sorted(files), however that sorts it the same as os.walk(path) already sorts them. If I try sorted(files, reverse=True), then while the originals come before the duplicates, multiple duplicates are now sorted backward and all the originals are backward too, ie:

IMG0003.jpg
IMG0002.jpg
IMG0002(2).jpg
IMG0002(1).jpg
IMG0001.jpg

String ordering is lexicographic, so you'll need a custom sort key if you want something different. It's a little trickier than expected, but something like this should work:

import os
import re

def key(fname):
    basename, ext = os.path.splitext(fname)
    v = 0
    if m := re.match(r"(.*)\((\d+)\)$", basename):
        basename, v = m.groups()
        v = int(v)
    return basename, ext, v

Now you should be able to use something like files.sort(key=key).

Using pathlib.Path for more awareness of filename semantics, build a tuple
with special cases as leading elements and the filename at the end. Sort that list of tuples but keep only the last element.

def test():
from pathlib import Path
def filenamesort(inp : list[str]):
"""build a list of custom tuples from the filename list, sort it and 
return the rightmost field, which is the filename.
"""
def tupleize(v):
""" returns a tuple of strings based on Path.stem for the filename
special case.  split into the part before the last `(` and what comes after
IMG0002(1).jpg => ('IMG0002', 1, 'IMG0002(2).jpg')
normal case, return the stem and an empty value
IMG0002.jpg  => ('IMG0002', 0, 'IMG0002.jpg')
The last element, least significant to sort is the filename
to be more solid foo(xxx).jpg should be ignored as xxx is not a numeric.
"""
pa = Path(v)
stem = pa.stem
if stem.endswith(")"):
lead, seq = stem.rsplit("(",maxsplit=1)
return (lead,int(seq.rstrip(")")),v)
else:
# "" will sort before "1)"
return (stem,0,v)
li = [tupleize(v) for v in inp]
#sort the list then return the last position in the tuple: the filename proper
return [v[-1] for v in sorted(li)]
def fmt(sin : str):
res = [v for line in sin.splitlines() if (v:=line.strip())]
return res
inp = fmt("""IMG0001.jpg
IMG0002(1).jpg
IMG0002(2).jpg
IMG0002(11).jpg
IMG0002.jpg
IMG0003.jpg
""")
exp = fmt("""IMG0001.jpg
IMG0002.jpg
IMG0002(1).jpg
IMG0002(2).jpg
IMG0002(11).jpg
IMG0003.jpg""")
dataexp = [
(inp,exp),
]
for inp, exp in dataexp:
for f in [filenamesort]:
got = f(inp)
msg = f"\n{f.__name__} for {str(inp):100.100} \nexp :{exp}:\ngot :{got}:\n"
if exp == got:
print(f"✅! {msg}")
else:
print(f"❌!  {msg}")
test()

output:

✅! 
filenamesort for ['IMG0001.jpg', 'IMG0002(1).jpg', 'IMG0002(2).jpg', 'IMG0002(11).jpg', 'IMG0002.jpg', 'IMG0003.jpg'] 
exp :['IMG0001.jpg', 'IMG0002.jpg', 'IMG0002(1).jpg', 'IMG0002(2).jpg', 'IMG0002(11).jpg', 'IMG0003.jpg']:
got :['IMG0001.jpg', 'IMG0002.jpg', 'IMG0002(1).jpg', 'IMG0002(2).jpg', 'IMG0002(11).jpg', 'IMG0003.jpg']:

I've checked that tupleize (the one for the tuple) can be used as a key parameter to sort.

i.e. this works as well sorted(inp,key=tupleize)

Wim was right, this was failing with different extension. Fixed with adjustment below:

        pa = Path(v)
stem = pa.stem
if stem.endswith(")"):
lead, seq = stem.rsplit("(",maxsplit=1)
return (lead,pa.suffix, int(seq.rstrip(")")),v)
else:
# "" will sort before "1)"
return (stem,pa.suffix,0,v)