Function to ensure a filename is valid

makeValidFilename.py

import os
import re
import platform

def makeValidFilename(value, normalize_unicode = False, windows_safe = False, custom_blacklist = None):
    """
    Takes a string and makes it into a valid filename.

    normalize_unicode replaces accented characters with ASCII equivalent, and
    removes characters that cannot be converted sensibly to ASCII.

    windows_safe forces Windows-safe filenames, regardless of current platform

    custom_blacklist specifies additional characters that will removed. This
    will not touch the extension separator:

        >>> makeValidFilename("T.est.avi", custom_blacklist=".")
        'T_est.avi'
    """

    if windows_safe:
        # Allow user to make Windows-safe filenames, if they so choose
        sysname = "Windows"
    else:
        sysname = platform.system()

    # Treat extension seperatly
    value, extension = os.path.splitext(value)

    # Remove null byte
    value = value.replace("\0", "")

    # If the filename starts with a . prepend it with an underscore, so it
    # doesn't become hidden
    if value.startswith("."):
        value = "_" + value

    # Blacklist of characters
    if sysname == 'Darwin':
        # : is technically allowed, but Finder will treat it as / and will
        # generally cause weird behaviour, so treat it as invalid.
        blacklist = r"/:"
    elif sysname == 'Linux':
        blacklist = r"/"
    else:
        # platform.system docs say it could also return "Windows" or "Java".
        # Failsafe and use Windows sanitisation for Java, as it could be any
        # operating system.
        blacklist = r"\/:*?\"<>|"

    # Append custom blacklisted characters
    if custom_blacklist is not None:
        blacklist += custom_blacklist

    # Replace every blacklisted character with a underscore
    value = re.sub("[%s]" % re.escape(blacklist), "_", value)

    # Remove any trailing whitespace
    value = value.strip()

    # There are a bunch of filenames that are not allowed on Windows.
    # As with character blacklist, treat non Darwin/Linux platforms as Windows
    if sysname not in ['Darwin', 'Linux']:
        invalid_filenames = ["CON", "PRN", "AUX", "NUL", "COM1", "COM2",
        "COM3", "COM4", "COM5", "COM6", "COM7", "COM8", "COM9", "LPT1",
        "LPT2", "LPT3", "LPT4", "LPT5", "LPT6", "LPT7", "LPT8", "LPT9"]
        if value in invalid_filenames:
            value = "_" + value

    # Replace accented characters with ASCII equivalent
    if normalize_unicode:
        import unicodedata
        value = unicode(value) # cast data to unicode
        value = unicodedata.normalize('NFKD', value).encode('ascii', 'ignore')

    # Truncate filenames to valid length
    if sysname in ['Darwin', 'Linux']:
        max_len = 255
    else:
        max_len = 32

    if len(value + extension) > max_len:
        if len(extension) > len(value):
            # Truncate extension instead of filename, no extension should be
            # this long..
            new_length = max_len - len(value)
            extension = extension[:new_length]
        else:
            new_length = max_len - len(extension)
            value = value[:new_length]

    return value + extension

def test():
    def assertEquals(a, b):
        assert a == b, "Error, %r not equal to %r" % (a, b)

    assertEquals(makeValidFilename("test.avi"), "test.avi")
    assertEquals(makeValidFilename("Test File.avi"), "Test File.avi")
    assertEquals(makeValidFilename("Test"), "Test")

    assertEquals(makeValidFilename("Test/File.avi"), "Test_File.avi")
    assertEquals(makeValidFilename("Test/File"), "Test_File")

    assertEquals(makeValidFilename("Test/File.avi", windows_safe = True), "Test_File.avi")
    assertEquals(makeValidFilename("\\/:*?<Evil>|\"", windows_safe = True), "______Evil___")
    assertEquals(makeValidFilename("COM2.txt", windows_safe = True), "_COM2.txt")
    assertEquals(makeValidFilename("COM2", windows_safe = True), "_COM2")

    assertEquals(makeValidFilename("."), "_.")
    assertEquals(makeValidFilename(".."), "_..")
    assertEquals(makeValidFilename("..."), "_...")

    assertEquals(makeValidFilename("Test.avi", custom_blacklist="e"), "T_st.avi")
        
    assertEquals(makeValidFilename("a" * 300), "a" * 255)
    assertEquals(makeValidFilename("a" * 255 + ".avi"), "a" * 251 + ".avi")
    assertEquals(makeValidFilename("a" * 251 + "b" * 10 + ".avi"), "a" * 251 + ".avi")
    assertEquals(makeValidFilename("test." + "a" * 255), "test." + "a" * 250)


if __name__ == '__main__':
    test()