A script used to recursively check a path for suspicious (less compatible with portable devices) filenames

check-path.py

#!/usr/bin/python

import os
import sys

# Paths longer than this amount of bytes will generate a warning
MAX_PATH_LEN = 256

# Good characters
GOOD_CHARS = " 0123456789abcdefghijklmnopqrstuvwxyzABCDEFGHIJKLMNOPQRSTUVWXYZ!?'._-()[]"


# Extensions whitelist/blacklist
EXT_WHITELIST = [ 'mp3', 'ogg', 'flac', 'wma', 'aac' ]
EXT_BLACKLIST = [ 'db' ]

# Generate a warning if non-whitelisted extensions present?
WARN_UNKNOWN_EXT = True

# Print notes?
PRINT_NOTES = True


def warn(path, notes=[]):
    """Prints a warning for a given path with all the extra notes, if enabled"""
    print "[W]", path
    if PRINT_NOTES:
        for note in notes:
            print "  *", note
    print


def err(path, notes=[]):
    """Prints an error for a given path with all the extra notes, if enabled"""
    print "[E]", path
    if PRINT_NOTES:
        for note in notes:
            print "  *", note
    print


def hilight_chars(s, idx=[]):
    output = list(s)
    for i in idx:
        output[i] = "<%s>" % output[i]
    return "".join(output)


def count_dots(name):
    return len(filter(lambda ch: ch == '.', list(name)))


def get_bad_chars(name):
    i = 0
    bad_char_idx = []
    for ch in name:
        if ch not in GOOD_CHARS:
            bad_char_idx += [i]
        i += 1
    return bad_char_idx


def get_non_ascii(name):
    i = 0
    bad_char_idx = []
    for ch in name:
        o = ord(ch)
        if o < 0x20 or o > 0x7E:
            bad_char_idx += [i]
        i += 1
    return bad_char_idx


def check_dirs(path, dirs):
    for d in dirs:
        warnings = []
        errors = []
        full_path = os.path.join(path, d)

        if d == "":
            error += ["WEIRD: EMPTY DIRECTORY NAME"]
            continue

        # Check how many dots does this have
        if count_dots(d) > 0:
            warnings += ["One or more period in directory name"]

        # Does it start with a period?
        if d[0] == '.':
            warnings += ["Directory starts with a period (.something)"]

        # Does it have non-good characters?
        non_good_idx = get_bad_chars(d)
        if non_good_idx != []:
            warnings += ["Potentially bad characters in name"]
            warnings += ["  -> "+ hilight_chars(d, non_good_idx)]

        # Does it have non-ASCII characters?
        non_ascii_idx = get_non_ascii(d)
        if non_ascii_idx != []:
            errors += ["Non-ASCII characters in name!"]
            errors += ["  -> "+ hilight_chars(d, non_ascii_idx)]

        # Check path length
        path_len = len(full_path)
        if path_len > MAX_PATH_LEN:
            warnings += ["Long path (%d > %d)" % (path_len, MAX_PATH_LEN)]

        # Print results
        if errors != []:
            err(full_path, errors)
        if warnings != []:
            warn(full_path, warnings)


def check_files(path, files):
    for f in files:
        warnings = []
        errors = []
        full_path = os.path.join(path, f)

        if f == "":
            error += ["WEIRD: EMPTY FILE NAME"]
            continue

        # Does it have an extension?
        num_dots = count_dots(f)
        if num_dots == 0:
            errors += ["No file extension!"]
            continue

        # Check if it has several dots
        if num_dots > 1:
            warnings += ["More than one period in file name"]

        # Does it start with a period?
        if f[0] == '.':
            warnings += ["File starts with a period (.something)"]

        ext = f.split('.')[-1].lower()

        # Check extension length
        if len(ext) != 3:
            warnings += ["File extension is not 3 characters - suspicious"]

        # Check whitelist/blacklist
        if ext not in EXT_WHITELIST:
            if ext in EXT_BLACKLIST:
                errors += ["File extension is blacklisted!"]
            else:
                warnings += ["File extension (%s) not in whitelist!" % ext]

        # Does it have non-good characters?
        non_good_idx = get_bad_chars(f)
        if non_good_idx != []:
            warnings += ["Potentially bad characters in name"]
            warnings += ["  -> "+ hilight_chars(f, non_good_idx)]

        # Does it have non-ASCII characters?
        non_ascii_idx = get_non_ascii(f)
        if non_ascii_idx != []:
            errors += ["Non-ASCII characters in name!"]
            errors += ["  -> "+ hilight_chars(f, non_ascii_idx)]

        # Check path length
        path_len = len(full_path)
        if path_len > MAX_PATH_LEN:
            warnings += ["Long path (%d > %d)" % (path_len, MAX_PATH_LEN)]

        # Print results
        if errors != []:
            err(full_path, errors)
        if warnings != []:
            warn(full_path, warnings)


def check(path, dirs, files):
    # Is it an empty dir?
    if dirs == [] and files == []:
        warn(path, ["Empty directory"])

    check_dirs(path, dirs)
    check_files(path, files)


def main(argv):
    if len(argv) > 1:
        path = argv[1]
    else:
        path = '.'

    # Ensure it exists
    if not os.path.exists(path):
        sys.stderr.write("Path '%s' does not exist\n")
        return 1

    # Walk through...
    for (pathname, dirs, files) in os.walk(path):
        check(pathname, dirs, files)

    # Done
    print
    print "DONE"
    return 0


###--# Initialization #--######################################################
if __name__ == '__main__':
    raise SystemExit( main(sys.argv) )