Находим группы дубликатов

def files_equal(path_a: Path, path_b: Path, chunk_size: int = DEFAULT_CHUNK_SIZE) -> bool:
    """Финальное подтверждение — совпадающий дайджест не гарантия."""
    with path_a.open("rb") as file_a, path_b.open("rb") as file_b:
        while True:
            chunk_a = file_a.read(chunk_size)
            chunk_b = file_b.read(chunk_size)
            if chunk_a != chunk_b:
                return False
            if not chunk_a:
                return True

def find_duplicates(files: list[FileInfo]) -> list[DuplicateGroup]:
    by_size = defaultdict(list)
    for file in files:
        by_size[file.size].append(file)

    groups = []
    for size, candidates in by_size.items():
        if len(candidates) < 2:
            continue
        by_digest = defaultdict(list)
        for candidate in candidates:
            digest = sha256_file(candidate.path)
            by_digest[digest].append(candidate)
        for digest, matched in by_digest.items():
            if len(matched) < 2:
                continue
            exact_groups = []
            for candidate in matched:
                for exact_group in exact_groups:
                    if files_equal(exact_group[0].path, candidate.path):
                        exact_group.append(candidate)
                        break
                else:
                    exact_groups.append([candidate])
            for exact_group in exact_groups:
                if len(exact_group) >= 2:
                    groups.append(DuplicateGroup(size=size, digest=digest, files=tuple(exact_group)))
    return groups

def test_equal_content_forms_one_group(tmp_path):
    a = tmp_path / "a.bin"
    b = tmp_path / "b.bin"
    a.write_bytes(b"same")
    b.write_bytes(b"same")

    groups = find_duplicates(scan(tmp_path, Config()))

    assert len(groups) == 1
    assert {item.path for item in groups[0].files} == {a, b}