compare two directories and optionally replace one's content (i.e., from a backup)

compare_and_replace.py

'''
compare two directories recursively and optionally
replace directories' contents with the other

use case: replacing files from an old backup for 
a cloned hard drive that has unpredictably corrupt files

'''

import os
import hashlib
from shutil import copy2


def md5(fname):
    hash_md5 = hashlib.md5()
    with open(fname, 'rb') as f:
        for chunk in iter(lambda: f.read(4096), b""):
            hash_md5.update(chunk)
    return hash_md5.hexdigest()


def comparedirs(goodpath, suspiciouspath, replace=None):
    ''' compares two directories' contents, returning lists
    of missing or corrupt (or changed) files. if replace is set, will
    replace some files from the goodpath. replace options: 'missing',
    'corrupt', 'both', or 'confirm'  '''
    # POSSIBLY TO DO -- CONSIDER ASSESSING 'EXTRA' FILES
    okfiles = []
    missingfiles = []
    corruptfiles = []
    directlength = len(goodpath.split(os.sep))
    for (root, dirs, files) in os.walk(goodpath):
        print('-' * 45)
        print(root)
        print(str(len(files)), 'files, ', str(len(dirs)), 
              'directories')
        for f in files:
            apath = root.split(os.sep)[directlength:]
            spath = os.path.join(suspiciouspath, (os.sep).join(apath))
            safefile = os.path.join(root, f)
            suspfile = os.path.join(spath, f)
            if os.path.exists(suspfile):
                x = md5(safefile)
                y = md5(suspfile)
                if x == y:
                    okfiles.append(suspfile)
                else:
                    corruptfiles.append(suspfile)
                    if replace in ['corrupt', 'both', 'confirm']:
                        replace, _ = confirmreplace(replace, f, 'is corrupt')
                        if _:
                            copy2(safefile, suspfile)
            else:
                missingfiles.append(suspfile)
                if replace in ['missing', 'both', 'confirm']:
                    replace, _ = confirmreplace(replace, f, 'is missing')
                    if _:
                        os.makedirs(spath, exist_ok=True)
                        copy2(safefile, suspfile)
    return okfiles, missingfiles, corruptfiles


def confirmreplace(replace, f, reason):
    if replace != 'confirm':
        print('replacing', f)
        return replace, True
    print('[enter "both", "corrupt", or "missing" to halt confirmations]')
    print(f, reason, end=' -- ')
    response = input('replace it? ').lower()
    if response == 'y':
        print('ok, replacing...')
        return replace, True
    elif response in ['both', 'corrupt', 'missing']:
        return response, True
    else:
        return replace, False

goodpath = r'C:\path\with\confirmed\files'
suspiciouspath = r'C:\location\of\dubious\files'

ok, missing, corrupt = comparedirs(goodpath, suspiciouspath, replace='confirm')