Filters Gor capture files and removes any requests bigger than 64KiB

filter_capture.py

#!/usr/bin/env python

from __future__ import print_function
from bitstring import ConstBitStream
import sys

import argparse

# Requests in gor file are demarked by  \n<U+1F435><U+1F648><U+1F649>\n
boundary = "0x0af09f90b5f09f9988f09f99890a"

# Max payload gor can read from a file is currently 64Kb
max_size = 64 * 1024 * 8


def verbose(*objs):
    print(*objs, file=sys.stderr)


def do_filter(instream, outstream):
    s = ConstBitStream(instream)

    discarded = 0

    try:
        while s.pos < s.len:
            bs = s.readto(boundary, bytealigned=True)

            if bs.length > max_size:
                discarded += 1
            else:
                outstream.write(bs.tobytes())
    finally:
        outstream.close()
        verbose("Discarded {} payloads".format(discarded))


if __name__ == '__main__':
    parser = argparse.ArgumentParser(description="Filters a gor capture file and removes any requests larger than 64Kb")
    parser.add_argument('-i', '--input-file', type=argparse.FileType('rb'), default='-', dest='instream')
    parser.add_argument('-o', '--output-file', type=argparse.FileType('wb'), default='-', dest='outstream')

    args = parser.parse_args()
    do_filter(args.instream, args.outstream)