adding rescue file splitter and some unit tests using it

[python-delta-tar] / filesplit.py

#!/usr/bin/env python

import argparse

BUFSIZE = 16 * 1024

def split_file(separator, prefix, input_file):
    '''
    splits a file when it finds a regexp, including the regexp in the begining
    of the new file
    '''
    i = 0
    pos = 0
    buf = ""
    sep_len = len(separator)
    if sep_len == 0:
        raise Exception("empty separator")

    output = open(prefix + str(i), 'w')

    # buffered search. we try not to have the while input file in memory, as
    # it's not needed
    with open(input_file, 'r') as f:
        while True:
            buf += f.read(BUFSIZE)
            if len(buf) == 0:
                break

            # split using the separator
            while separator in buf:
                idx = buf.index(separator)

                if idx > 0:
                    output.write(buf[0:idx])
                    output.close()
                    i += 1
                    output = open(prefix + str(i), 'w')
                    output.write(buf[idx:idx + sep_len])
                else:
                    output.write(buf[0:sep_len])

                buf = buf[idx + sep_len:]

            # corner case: separator is between this buf and next one. In this
            # case, we write to current output everything before that and
            # iterate
            if separator[0] in buf[-sep_len:]:
                output.write(buf[:-sep_len])
                buf = buf[-sep_len:]
                continue

            # else: continue writing to the current output and iterate
            output.write(buf)
            buf = ""

    output.close()

if __name__ == "__main__":
    parser = argparse.ArgumentParser()

    parser.add_argument("-s", "--separator", required=True,
                        help="string for the separator")
    parser.add_argument("-p", "--prefix", required=True,
                        help="prefix for split files")
    parser.add_argument("input_file", help="input file")

    args = parser.parse_args()
    split_file(separator=args.separator, prefix=args.prefix, input_file=args.input_file)