70 lines
2.2 KiB
Python

import argparse
import logging
from pathlib import Path
from tqdm import tqdm
from karaokatalog.deduplicate.find_duplicates import find_duplicates
from karaokatalog.deduplicate.prune import prune
from karaokatalog.Library import Library
from karaokatalog.Song import Song
from karaokatalog.util.get_equivalence_classes import get_equivalence_classes
logging.basicConfig(
format="%(asctime)s [%(levelname)s] %(message)s", level=logging.INFO
)
def parse_args() -> argparse.Namespace:
parser = argparse.ArgumentParser(
prog="python3 -m karaokatalog.deduplicate",
description="Deduplicate UltraStar Deluxe song libraries",
)
parser.add_argument(
"library_path",
type=Path,
help="The directory which contains the songs, the one you'd also configure UltraStar Deluxe to use",
)
return parser.parse_args()
if __name__ == "__main__":
args = parse_args()
logging.info("Karaokatalog Deduplication started")
logging.info("Loading library")
library = Library.from_dir(args.library_path)
logging.info("Library loaded")
logging.info("Finding duplicates (songs with identical title and artist)")
duplicates = find_duplicates(library.songs_in_unique_dirs)
logging.info("Duplicates found")
logging.info("Finding exact duplicates (files are 100% identical)")
# TODO Make this abortable and restartable
exact_duplicates = tuple(
equivalence_class
for songs in tqdm(duplicates.values(), unit=" duplicates")
for equivalence_class in get_equivalence_classes(songs, Song.has_identic_files)
)
logging.info("Exact duplicates found")
logging.info("Determining songs to prune")
pruning_instructions = tuple(
instruction
for equivalence_class in tqdm(exact_duplicates, unit=" exact duplicates")
for instruction in prune(equivalence_class)
)
logging.info(
f"{len(pruning_instructions)} exactly duplicated songs will be deleted"
)
logging.warning(f"Deleting {len(pruning_instructions)} songs!")
for instruction in tqdm(pruning_instructions, unit=" songs"):
instruction()
logging.info("Deletion done")
logging.info("Karaokatalog Deduplication done")