How to define NamedTuple class containing a NumPy array

Question

I am trying to define a a class of namedtuples that contains an array, but with no success.

I have tried using np.zeros(shape=(16)) or np.empty(2, dtype = int8) but with no success.

from typing import NamedTuple
import numpy as np

class BasicDemodStatsMsg(NamedTuple):

    header: IcdHeader

    field1: np.uint32

    field2: np.uint32

    field3: np.uint16

    reserved: np.zeros(shape=(16))

    checksum: np.uint16

At the end I would like to have a class that has a fixed size of bytes in the memory and can be packed and unpacked to and from byte arrays.

Have you successfully defined such a class that does not contain an array? — mkrieger1
– mkrieger1, Commented Aug 29, 2019 at 8:31
It's collections.namedtuple not collections.NamedTuple @Yarom. — Ben
– Ben, Commented Aug 29, 2019 at 8:40
As @Ben said collections.namedtuple is a thing. The only NamedTuple I can think of right now is typing.NamedTuple and that is only for typing purpose. — Adirio
– Adirio, Commented Aug 29, 2019 at 8:42

Ben · Accepted Answer · 2019-08-29 12:39:36Z

3

Your issue is related to the data type; np.zeros() is a method which returns np.ndarray, the actual data type. This is fine:

>>> class BasicDemodStatsMsg(typing.NamedTuple):
...     field1: np.uint32
...     field2: np.uint32
...     field3: np.uint16
...     checksum: np.uint16
...     reserved: np.ndarray = np.zeros(shape=16)
...
>>>

You should be extremely wary about how reserved is used, as it's a mutable default argument. It may be more appropriate to use a dataclass and a therefore allow yourself to use the default factory to create it.

from dataclasses import dataclass, field
from typing import NamedTuple

@dataclass
class BasicDemodStatsMsg(NamedTuple):
    field1: np.uint32
    field2: np.uint32
    field3: np.uint16
    checksum: np.uint16
    reserved: np.ndarray = field(default_factory=lambda: np.zeros(shape=16, dtype=np.uint8))

Note that the default attribute is the last one declared; this is because non-default fields can't follow defaults in named tuples.

edited Aug 29, 2019 at 12:39

answered Aug 29, 2019 at 8:44

Ben

53.1k36 gold badges133 silver badges156 bronze badges

Sign up to request clarification or add additional context in comments.

2 Comments

Yarom Swissa Over a year ago

i need to specify the type of the array, for example uint8, which does not happen in your solution.

Ben Over a year ago

You can just specify the type normally @Yarom: reserved: np.ndarray = field(default_factory=lambda: np.zeros(shape=16, dtype=np.uint8)) I've also updated the answer to demonstrate this.

Collectives™ on Stack Overflow

How to define NamedTuple class containing a NumPy array

1 Answer 1

2 Comments

Your Answer

Linked

Hot Network Questions

Collectives™ on Stack Overflow

1 Answer 1

2 Comments

Your Answer

Sign up or log in

Post as a guest

Linked

Related