dedup

#!/usr/bin/env python3
"""
Move duplicates into a common folder for easy deletion.
"""
from pathlib import Path
from filecmp import cmp

dup_folder = Path("_dups")


def move_if_same(f1, f2):
    """Moves f2 if it's identical to f1."""
    assert f1.is_file()
    assert f2.is_file()
    dup_folder.mkdir(exist_ok=True)
    # cmp uses byte-by-byte comparison in shallow=False mode.
    if cmp(str(f1.absolute()), str(f2.absolute()), shallow=False):
        print(f"Moving duplicate '{f2}', which is the same as '{f1}'.")
        # Raises if targets exists.
        f2.rename(dup_folder / f2)
        return True

    return False


def main():
    files = sorted(p for p in Path(".").glob("*") if p.is_file())

    deleted = set()
    for i in range(len(files)):
        if i in deleted:
            continue
        for j in range(i + 1, len(files)):
            if j in deleted:
                continue
            if move_if_same(files[i], files[j]):
                deleted.add(j)


if __name__ == "__main__":
    main()