Compare two files containing (default) md5sum output and print results.

md5sum-diff

#!/usr/bin/env python3
#
# Copyright (C) 2018  Jon Szymaniak <[email protected]>
# SPDX License Identifier: MIT
#
# Compare two files containing (default) md5sum output and print results.
#
################################################################################

__version__ = '0.1.1'

import os
import re
import sys

class MD5SumResults:
    _re = re.compile('(?P<hash>[a-fA-F0-9]{32})\s+(?P<file>.*)$')

    def __init__(self, filename):
        '''
        Load the md5sum results contained in `filename`
        '''
        self._hashes = {}
        self.filename = filename

        with open(filename, 'r') as infile:
            for line in infile:
                if len(line) != 0:
                    m = MD5SumResults._re.match(line)
                    if m:
                        self._hashes[m['hash']] = m['file']

    def match(self, other):
        """
        Returns a list of matching hashes. Each list entry is a tuple in the form:
            (hash, this object's filename, other object's filename)
        """
        ret = []
        for target in self._hashes:
            for entry in other._hashes:
                if entry == target:
                    ret.append((target, self._hashes[target], other._hashes[target]))
                    break
        return ret

    def no_match(self, other):
        """
        Returns a list of hashes that occur in this object, but not in `other`.
        Each list entry is in the form:
            (hash, filename)
        """
        ret = []
        for target in self._hashes:
            match = False
            for entry in other._hashes:
                if entry == target:
                    match = True
                    break

            if not match:
                ret.append((target, self._hashes[target]))
        return ret


if __name__ == "__main__":
    argv_len = len(sys.argv)
    show_same = True
    show_diff = True

    if argv_len < 3 or argv_len > 4:
        print('md5sum-diff ({:s})'.format(__version__))
        print('Usage: {:s} <file1> <file2> [same|diff]'.format(os.path.basename(sys.argv[0])))
        print('Compare two files containing (default) md5sum output and print results.')
        print('The optional last argument specifies whether the matches or differences are printed.')
        sys.exit(0)

    if argv_len == 4:
        if sys.argv[3] == 'same':
            show_diff = False
        elif sys.argv[3] == 'diff':
            show_same = False
        else:
            print('Invalid argument: {:s}'.format(argv[4]), file=sys.stderr)
            sys.exit(1)


    a = MD5SumResults(sys.argv[1])
    b = MD5SumResults(sys.argv[2])

    matches = a.match(b)    if show_same else []
    a_diffs = a.no_match(b) if show_diff else []
    b_diffs = b.no_match(a) if show_diff else []

    if len(matches) != 0:
        print('\nMatching Hashes')
        print('-' * 32)
        for entry in matches:
            print('{:34s}{:32s}{:32s}'.format(entry[0], entry[1], entry[2]))

    if len(a_diffs) != 0:
        print('\nIn {:s} but not in {:s}'.format(a.filename, b.filename))
        print('-' * 32)
        for entry in a_diffs:
            print('{:34s}{:32s}'.format(entry[0], entry[1]))

    if len(b_diffs) != 0:
        print('\nIn {:s} but not in {:s}'.format(b.filename, a.filename))
        print('-' * 32)
        for entry in b_diffs:
            print('{:34s}{:32s}'.format(entry[0], entry[1]))

    print()