catb

#!/usr/bin/env python
"""
catb: compare files by printing a column of bytes per file side-by-side
author: graemeball@googlemail.com, 2013
license: GPL
"""

import sys


NARGS = len(sys.argv)
if NARGS < 2:
    print("\n  Usage: catb <filename1> <filename2> ... <filenameN>")
    print("\n    prints a byte column for each file, side-by-side")
    sys.exit(0)


def main():
    """
    For each file specified, print a stream of bytes in hex formatted
    columns, pausing when a difference between the files is found
    """
    files = []
    for arg_no in [file_no + 1 for file_no in range(NARGS - 1)]:
        files.append(open(sys.argv[arg_no], 'rb'))

    nbytes = 0
    file_bytes = read_bytes(files)
    while any(file_bytes):
        byte_str = "  {0:10d}:   ".format(nbytes)
        nbytes += 1
        byte_str += "    ".join([byte_string(b) for b in file_bytes])
        if len(set(file_bytes)) == 1:
            print(byte_str)
        else:
            print(byte_str + "    <= different")
            #raw_input("...Enter to continue")  # uncomment to pause upon diff
        file_bytes = read_bytes(files)

    for bfile in files:
        bfile.close()

    print("\ntotal nbytes: {0}".format(nbytes))


def byte_string(byte_val):
    """
    Return string representation of byte, or N! for length 0 string
    """
    if byte_val == "":
        return "N!"
    else:
        return "{0:x}".format(ord(byte_val))


def read_bytes(files):
    """
    Return a list of the next byte for each file in a file list
    """
    file_bytes = []
    for bfile in files:
        a_byte = bfile.read(1)
        file_bytes.append(a_byte)
    return file_bytes


if __name__ == '__main__':
    main()