Recursively list files within a directory

Question

I have the following code to recursively list files within a directory:

from fnmatch import filter
from os import path, walk
def files_within(directory_path):
 files_within = []
 for root, directory_path, files_path in walk(directory_path):
 for file_path in filter(files_path, "*"):
 files_within.append(path.join(root, file_path))
 return files_within

I want change it to avoid side affects. Is it possible?

I'm not using the glob because I'm using Python 2.5 because of a legacy code.

Answer 1

Your function could be a generator (which were introduced in Python 2.2). It should also expose the pattern for the filter as a parameter:

import os
import fnmatch
def files_within(directory_path, pattern="*"):
 for dirpath, dirnames, filenames in os.walk(directory_path):
 for file_name in fnmatch.filter(filenames, pattern):
 yield os.path.join(dirpath, file_name)
if __name__ == "__main__":
 # You can simply iterate over the generator:
 for file_path in files_within("."):
 print file_path
 # If you absolutely need a list:
 file_list = list(files_within("../"))
 # Use the pattern parameter:
 text_files = list(files_within(".", "*.txt"))

I like to leave the names from os.walk the same as in the documentation, but this is a personal choice.

What is not quite a personal choice is the number of tabs per indentation level. Python's official style-guide, PEP8, recommends using always 4 spaces.

I also changed the way of the imports. Here it already becomes obvious why importing them like this is better (as far as readability goes). It is obvious that fnmatch.filter is from the module fnmatch and not the built-in filter.

• Graipher: @FátimaAlves You're welcome. Even though I'm still not entirely sure what you mean by "side effects" in your question. Was it the overriding of the function name with the internal local variable?
• FXux: I mean avoid reassign variables already assigned.