Find duplicate music files and empty dirs

findDuplicateMusic.py

#!/usr/bin/env python


"""Look for "bad" directories with duplicate files (xxx.mp3 vs xxx 1.mp3),
empties, only artwork, etc
"""


import os
import re
import sys
from send2trash import send2trash


def folder_size(path='.'):
    total = 0
    for entry in os.scandir(path):
        if entry.is_file():
            total += entry.stat().st_size
        elif entry.is_dir():
            total += folder_size(entry.path)
    return total


def filter_useless(contents):
    # FIXME really should check final filename, but cant image a dir named that...
    return [c for c in contents if ".DS_Store" not in c]


AUDIO_FMTS = [".mp3", ".wav", ".wma", ".m4a", ".ogg", ".flac"]


# If True, send to trash immediately, otherwise just write txtfile with names
DELETE_NOW = True


def send_to_trash(filename):
    if DELETE_NOW:
        send2trash(filename)


to_delete = []
total_size = 0
for dirname in os.listdir('.'):
    if not os.path.isdir(dirname):
        continue

    dir_contents = filter_useless(os.listdir(dirname))

    if len(dir_contents) == 0:
        print('*** Empty:', dirname)
        to_delete.append(dirname)
        send_to_trash(dirname)
        continue

    for album in dir_contents:
        full_album_path = os.path.join(dirname, album)
        if not os.path.isdir(full_album_path):
            continue

        contents = filter_useless(os.listdir(full_album_path))
        if len(contents) == 0:
            print('*** Empty:', full_album_path)
            to_delete.append(full_album_path)
            send_to_trash(full_album_path)
            continue

        found_audio = any(os.path.splitext(fn)[1] in AUDIO_FMTS for fn in contents)

        if not found_audio:
            print("*** No music in dir:", full_album_path)
            to_delete.append(full_album_path)
            total_size += folder_size(full_album_path)
            send_to_trash(full_album_path)

        for filename in contents:
            stem, ext = os.path.splitext(filename)
            # re.escape vital for filenames with brackets, hypens, etc
            dupe_pattern = re.compile(re.escape(stem) +r' [0-9]' + ext)
            for fn in contents:
                if re.match(dupe_pattern, fn):
                    print('*** Original:', filename, 'Duplicate:', fn)
                    full_filename = os.path.join(full_album_path, fn)
                    to_delete.append(full_filename)
                    total_size += os.path.getsize(full_filename)
                    send_to_trash(full_filename)


print("You will save %.1f MB" % (total_size / (1024*1024)))
if not DELETE_NOW:
    with open('toDelete.txt', 'w') as of:
        for fname in to_delete:
            of.write(fname + "\n")