Find all files ending in lots of null bytes

null-bytes.py

#!/usr/bin/env python3

"""
This script recurses through the current user's home directory and prints the
paths to all files that end in 2 kB or more of null bytes.

The motivation for this is that Time Machine seems to have corrupted my music
files. Several kilobytes' worth of data have gone missing from the middle of
some files and replaced with an equivalent number of null bytes at the end.
"""

import os

skipdirs = ['node_modules', '.virtualenv', '__pycache__', '.git', 'Library']
skipfiles = ['.sqlite', '.sock', '.db', '.DS_store']
# How many bytes to check at the end of the file
byte_count = 2000
home_dir = os.path.expanduser('~')

for dirpath, dirs, files in os.walk(home_dir):
    if any([d in dirpath for d in skipdirs]):
        continue

    for file in files:
        null_byte_count = 0
        if any([f in file for f in skipfiles]):
            continue

        filepath = dirpath + '/' + file

        try:
            with open(filepath, 'rb') as f:
                try:
                    f.seek(-byte_count, 2)
                    last_bytes = f.read(byte_count)
                    if all(byte == 0 for byte in last_bytes):
                        print(filepath)
                except OSError as err:
                    continue

        except FileNotFoundError as err:
            pass
        except OSError as err:
            pass