jilorg_fix_filenames.py

#!/usr/bin/env python
"""
Fixes file names for downloads from jil.org.

jil.org delivers downloaded files with file names in e-mail header format. The
fix_filename function from this module can decode such file names.

When this module is executed as main program, all file names passed as program
arguments are renamed properly:

$ jilorg_fix_filenames.py =?UTF-8?B?bXl3aWRnZXQud2d0?= path/to/=?UTF-8?B?bXl3aWRnZXQud2d0?=
Moves "=?UTF-8?B?bXl3aWRnZXQud2d0?=" to "mywidget.wgt" and
"path/to/=?UTF-8?B?bXl3aWRnZXQud2d0?=" to "path/to/mywidget.wgt"

The easiest way is to use shell wildcards:
$ jilorg_fix_filenames.py path/to/downloads/=*
"""

from os.path import basename, dirname, join
from email.header import decode_header

def fix_filename(file_path):
    """
    Fixes filenames that are encoded as e-mail header.

    Also removes suffixes added by download software.

    >>> fix_filename("=?UTF-8?B?bXl3aWRnZXQud2d0?=")
    u'mywidget.wgt'

    >>> fix_filename("path/to/=?UTF-8?B?bXl3aWRnZXQud2d0?=")
    u'path/to/mywidget.wgt'

    >>> fix_filename("=?UTF-8?B?bXl3aWRnZXQud2d0?= (27)")
    u'mywidget.wgt'
    """
    cleaned = file_path[:file_path.rindex("=")+1]

    file_name = basename(cleaned)
    decoded = decode_header(file_name)[0]
    file_name = decoded[0].decode(decoded[1])

    dir_name = dirname(file_path)

    return join(dir_name, file_name)


if __name__ == "__main__":
    from sys import argv
    from shutil import move

    for path in argv[1:]:
        move(path, fix_filename(path))