Function ndependencies

dask/order.py:705–751

Number of total data elements on which this key depends For each key we return the number of tasks that must be run for us to run this task. Examples -------- >>> inc = lambda x: x + 1 >>> dsk = {'a': 1, 'b': (inc, 'a'), 'c': (inc, 'b')} >>> dependencies, dependents = g

(
    dependencies: Mapping[Key, set[Key]], dependents: Mapping[Key, set[Key]]
)

def ndependencies(
    dependencies: Mapping[Key, set[Key]], dependents: Mapping[Key, set[Key]]
) -> tuple[dict[Key, int], dict[Key, int]]:
    """Number of total data elements on which this key depends

    For each key we return the number of tasks that must be run for us to run
    this task.

    Examples
    --------
    >>> inc = lambda x: x + 1
    >>> dsk = {'a': 1, 'b': (inc, 'a'), 'c': (inc, 'b')}
    >>> dependencies, dependents = get_deps(dsk)
    >>> num_dependencies, total_dependencies = ndependencies(dependencies, dependents)
    >>> sorted(total_dependencies.items())
    [('a', 1), ('b', 2), ('c', 3)]

    Returns
    -------
    num_dependencies: Dict[key, int]
    total_dependencies: Dict[key, int]
    """
    num_needed = {}
    result = {}
    for k, v in dependencies.items():
        num_needed[k] = len(v)
        if not v:
            result[k] = 1

    num_dependencies = num_needed.copy()
    current: list[Key] = []
    current_pop = current.pop
    current_append = current.append

    for key in result:
        for parent in dependents[key]:
            num_needed[parent] -= 1
            if not num_needed[parent]:
                current_append(parent)
    while current:
        key = current_pop()
        result[key] = 1 + sum(result[child] for child in dependencies[key])
        for parent in dependents[key]:
            num_needed[parent] -= 1
            if not num_needed[parent]:
                current_append(parent)
    return num_dependencies, result


OrderInfo = namedtuple(