Accessing a nested class from another nested dataclass

Question

Now the following code works perfectly with Python 3.7

class A:
    class B:
        def __init__(self):
            print("B")

    class C:
        def __init__(self):
            self.b = A.B()

def main():
    a = A.C()

if __name__ == "__main__":
    main()

It prints a B on the screen.

However, with a small modification that tries to introduce dataclass, the code cannot run well.

from dataclasses import dataclass

class A:
    class B:
        def __init__(self):
            print("B")

    @dataclass
    class C:
        b = A.B()

def main():
    a = A.C()

if __name__ == "__main__":
    main()

Python reports -- for b = A.B() -- NameError: name 'A' is not defined.

Does anyone know how to fix this issue to achieve the same result with dataclass? And why does it say name 'A' is not defined?

Just unnest your class, and if you really need that namespacing, just add A.C = C. Note, your dataclass is not equivalent to what you had before. — juanpa.arrivillaga
– juanpa.arrivillaga, Commented Oct 2, 2019 at 5:15

blhsing · Accepted Answer · 2019-10-02 04:41:47Z

4

A class object is not created until the end of body of a class statement is reached, which is why your class A cannot be referenced while it is still being defined. Referencing A inside the __init__ method, on the other hand, is valid because the class A is already defined when the __init__ method is called.

You can instead use typing.TypeVar to define a forward-referencing type A.B for b, and assign to it a default value of a field with a default_factory function that returns an instance of A.B when called:

from dataclasses import dataclass, field
from typing import TypeVar

class A:
    class B:
        def __init__(self):
            print("B")

    @dataclass
    class C:
        b: TypeVar('A.B') = field(default_factory=lambda: A.B())

def main():
    a = A.C()

if __name__ == "__main__":
    main()

This outputs:

answered Oct 2, 2019 at 4:41

blhsing

109k9 gold badges88 silver badges132 bronze badges

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

3 Comments

aafulei Over a year ago

Thanks for the explanation. Can I ask for some clarification: What do you mean by a class is being defined? In both programs, C should be part of A. Why is it that calling __init__ method of C in the first program is thought to be done after A's definition, while specifying b = A.B() in C in the second program is thought to be done within A's definition?

Pynchia Over a year ago

class is an instruction in python. It's not a declaration like in other languages

aafulei Over a year ago

Thanks! But frankly I am still (or maybe even more) confused. Then what's the difference between an instruction and a declaration in this context? I have background in C/C++. If you would like to draw any comparison I am more than happy to hear.

Pynchia · Accepted Answer · 2019-10-02 05:32:41Z

0

Another way to defer its initialisation until class A is defined and available, without touching the generated __init__:

@dataclass
class C:
    b: TypeVar('A.B') = field(init=False)

def __post_init__(self):
    self.b = A.B()

Please refer to the official docs

edited Oct 2, 2019 at 5:32

answered Oct 2, 2019 at 5:16

Pynchia

11.7k5 gold badges38 silver badges49 bronze badges

1 Comment

TMoore Over a year ago

This is good unless you're defining C.b in the auto-generated __init__, i.e. C(b=A.B(foo)).

Collectives™ on Stack Overflow

Accessing a nested class from another nested dataclass

2 Answers 2

3 Comments

1 Comment

Your Answer

Hot Network Questions

Collectives™ on Stack Overflow

2 Answers 2

3 Comments

1 Comment

Your Answer

Sign up or log in

Post as a guest

Related