trim_whitespace.py

#!/usr/bin/env python

from __future__ import print_function, unicode_literals
from tempfile import SpooledTemporaryFile as TempFile
from shutil import copyfileobj
from argparse import ArgumentParser
from io import open


def trim_white(istr):
    trailing_empty_lines = 0

    for line in istr:
        line = line.rstrip()
        if line:
            yield "{0}{1}\n".format('\n' * trailing_empty_lines, line)
            trailing_empty_lines = 0
        else:
            trailing_empty_lines += 1


def main():
    parser = ArgumentParser()
    parser.add_argument('input_file')
    parser.add_argument('output_file', nargs='?', default=None)
    parser.add_argument('-e', '--encoding', default='utf-8')
    parser.add_argument('-m', '--memory', default=1024 * 1024, type=int,
        help="The maximum amount of memory to use as a buffer, before writing "
             "to a temporary file")

    args = parser.parse_args()

    input_file = args.input_file
    output_file = args.output_file or input_file
    encoding = args.encoding
    memory = args.memory

    with TempFile(max_size=memory, mode='w+b') as stage:
        with open(input_file, mode='rt', encoding=encoding) as istr:
            stage.writelines(line.encode(encoding) for line in trim_white(istr))

        stage.seek(0)

        with open(output_file, 'wb') as ostr:
            copyfileobj(stage, ostr)

if __name__ == '__main__':
    main()