Function _assert_diffs_match

python/pathway/tests/test_persistence_iterate.py:320–337

Assert the CSV output matches expected diffs exactly. The CSV has columns like (event_time, data, chunk_start, time, diff). We consolidate, drop time, and combine remaining columns into strings. Diffs must be exactly +1 or -1.

(output_path: pathlib.Path, expected: set)

def _assert_diffs_match(output_path: pathlib.Path, expected: set) -> None:
    """Assert the CSV output matches expected diffs exactly.

    The CSV has columns like (event_time, data, chunk_start, time, diff).
    We consolidate, drop time, and combine remaining columns into strings.
    Diffs must be exactly +1 or -1.
    """
    _assert_diffs_are_unit(output_path)
    try:
        df = consolidate(pd.read_csv(output_path))
    except pd.errors.EmptyDataError:
        assert expected == set(), f"Expected {expected} but output was empty"
        return

    actual = set(combine_columns(df))
    assert (
        actual == expected
    ), f"Diffs mismatch:\n  actual:   {sorted(actual)}\n  expected: {sorted(expected)}"


def _compute_chunk_assignments(events: dict) -> dict: