Proper use of dataclass decorator

Hello everyone,

I'm trying to create my own data class utilizing the @dataclass decorator.
Variables that I directly give values end up initialized propperly.
However the variables satisfied and loud keep their default values.

How could I force python to recalculate satisfied and loud values by given arguments?
I have attached a small code and its output representing my issue.

Code:

from dataclasses import dataclass, astuple

class BaseSatisfaction():
    satisfied = list[bool]
    loud = bool

@dataclass
class ClientSatisfaction(BaseSatisfaction):
    food_tasty: bool = False
    served_quickly: bool = False
    restaurant_noisiness_dB: None | float = None

    satisfied = [food_tasty, served_quickly]
    loud = 0 if restaurant_noisiness_dB is None else restaurant_noisiness_dB > 100
    
happy_client = ClientSatisfaction(food_tasty=True, served_quickly = True)
rage_client = ClientSatisfaction(restaurant_noisiness_dB = 120)

print("======== Happy Client =========")
print(happy_client.__dict__.items())
print(f"Happy clinet is satisfied: {happy_client.satisfied}")
print(f"How loud was it?: {happy_client.loud}")
print("======== Rage Client =========")
print(rage_client.__dict__.items())
print(f"Rage clinet is satisfied: {rage_client.satisfied}")
print(f"How loud was it?: {rage_client.loud}")

Output:

======== Happy Client =========
dict_items([('food_tasty', True), ('served_quickly', True), ('restaurant_noisiness_dB', None)]) 
Happy clinet is satisfied: [False, False]
Was it too loud?: 0
======== Rage Client =========
dict_items([('food_tasty', False), ('served_quickly', False), ('restaurant_noisiness_dB', 120)])
Rage clinet is satisfied: [False, False]
Was it too loud?: 0

It should return:

Happy clinet is satisfied: [True, True]

In Python, the code in the body of the class is executed only once, when the class is defined. This includes all assignments. This means that the assignment to satisfied and loud is performed when the class is defined, and uses the default values defined in the class body. This is why the initial values are not updated when an instance is created.

To solve this problem, you should compute the values of satisfied and loud for each instance. You can use the post_init method, which is a special method provided by the @dataclass decorator that is automatically called after the instance has been initialized.

Here is a fixed version of your code:

from dataclasses import dataclass, astuple

@dataclass
class ClientSatisfaction:
    food_tasty: bool = False
    served_quickly: bool = False
    restaurant_noisiness_dB: None | float = None

    def __post_init__(self):
        self.satisfied = [self.food_tasty, self.served_quickly]
        self.loud = 0 if self.restaurant_noisiness_dB is None else self.restaurant_noisiness_dB > 100

happy_client = ClientSatisfaction(food_tasty=True, served_quickly=True)
rage_client = ClientSatisfaction(restaurant_noisiness_dB=120)

print("======== Happy Client =========")
print(happy_client.__dict__.items())
print(f"Happy client is satisfied: {happy_client.satisfied}")
print(f"How loud was it?: {happy_client.loud}")
print("======== Rage Client =========")
print(rage_client.__dict__.items())
print(f"Rage client is satisfied: {rage_client.satisfied}")
print(f"How loud was it?: {rage_client.loud}")

With this code, the satisfied and loud fields are calculated for each instance, using the instance's own field values. This gives you the expected results. Note that you do not need the BaseSatisfaction class in this case.

As Pierre Vermot said above, these values are only set on initialization.

Since you apparently want to dynamically generate that attribute as a function of the two others (which might be altered after their initialization), I'd implement .satisfied as a @property: That way, it will at any time compute the satisfaction from the current values, like a function but accessed like an attribute:

Remove the satisfied attribute from the dataclass, and add this method instead:

@property
def satisfied(self):
   return [self.food_tasty, self.served_quickly]

Using it:

>>> satis = ClientSatisfaction(True, True)
>>> satis.satisfied
[True, True]
>>> satis.food_tasty = False
>>> satis.satisfied
[False, True]