detox-fn.py

#!/usr/bin/python3
# -*- coding: utf-8 -*-
# detox filenames so as to across most file systems
# (c) Copyright 2016 by Joseph Reagle
# Licensed under the GPLv3, see <http://www.gnu.org/licenses/gpl-3.0.html>
# 

import codecs
import glob
import locale
import logging
import os
from os import chdir, environ, mkdir, rename, walk
from os.path import abspath, basename, exists, isdir, isfile, splitext
import re
import sys
import unicodedata 

HOME = environ['HOME']

critical = logging.critical
info = logging.info
dbg = logging.debug
warn = logging.warn
error = logging.error
excpt = logging.exception

def remove_accents(s): 
    """remove accents for ascii analogs"""
    nkfd_form = unicodedata.normalize('NFKD', s) 
    return u''.join([c for c in nkfd_form if not unicodedata.combining(c)])

def remove_reserved(s):
    """only keep characters liked across most file systems"""
    # http://serverfault.com/questions/124611/special-characters-in-samba-filenames
    # https://amigotechnotes.wordpress.com/2015/04/02/invalid-characters-in-file-names/
    PERMITTED_CHARS = '''[^\w\-_\. \^&'@\{\}\[\],$=\!-#\(\)%.+~]'''
    s = s.replace(u'’', "'") # I like apostrophes, but no smart ones
    s = s.replace(u'?', "") # makes more sense just to remove '?'
    s = re.sub(PERMITTED_CHARS, '_', s)
    return s

def detoxify_fn(path, fn):
    new_fn = remove_accents(fn)
    new_fn = remove_reserved(new_fn)
    fn = os.path.join(path, fn)
    new_fn = os.path.join(path, new_fn)
    if new_fn != fn:
        try:
            print("detoxing '%s' to '%s'" %(fn, new_fn))
            if not args.no_rename:
                os.rename(fn, new_fn)
        except Exception as e:
            print(e)
    else:
        info('no need to rename %s' %fn)

def recurse(directory):
    for path, dirs, files in walk(directory):
        for fn in files:
            info("fn = %s" %fn)
            detoxify_fn(path, fn)

if '__main__' == __name__:

    import argparse # http://docs.python.org/dev/library/argparse.html
    arg_parser = argparse.ArgumentParser(description='detoxify filenames so as to work over most file systems')
    
    # positional arguments
    arg_parser.add_argument('files', nargs='+', metavar='FILE')
    # optional arguments
    arg_parser.add_argument("-n", "--no-rename",
        action="store_true", default=False,
        help="no renaming; perform dry run")
    arg_parser.add_argument('-L', '--log-to-file',
        action="store_true", default=False,
        help="log to file %(prog)s.log")
    arg_parser.add_argument('-V', '--verbose', action='count', default=0,
        help="Increase verbosity (specify multiple times for more)")
    arg_parser.add_argument('--version', action='version', version='TBD')
    args = arg_parser.parse_args()

    log_level = 100 # default
    if args.verbose == 1: log_level = logging.CRITICAL  #50
    elif args.verbose == 2: log_level = logging.INFO    #20
    elif args.verbose >= 3: log_level = logging.DEBUG   #10
    LOG_FORMAT = "%(levelno)s %(funcName).5s: %(message)s"
    if args.log_to_file:
        logging.basicConfig(filename='PROG-TEMPLATE.log', filemode='w',
            level=log_level, format = LOG_FORMAT)
    else:
        logging.basicConfig(level=log_level, format = LOG_FORMAT)

    if len(args.files) > 1:
        print("Sorry, I only take a single argument: "
            "a directory to recurse on or filename.")
        sys.exit()
    fn = args.files[0]
    if isfile(fn):
        detoxify_fn(fn)
    elif isdir(fn):
        recurse(fn)
    else:
        print("Sorry, argument is unknown file type")
        sys.exit()