1

In Matlab, let's consider a structure "structure" and a list of fields "list_of_fields".

structure = struct;
list_of_fields = ["a","b","d"]; % Some specific fields of the structure and not all of them
for field = list_of_field
    structure.(field) = 0;
end

I would like to do the same in Python with the attribute of a class: instead of writing:

class Example
    def __init__(self) :
        self.a = 0
        self.b = 0
        self.d = 0

My goal would be to write a loop that iterates over the specific attribute a, b and d (and no other attribute). How do you achieve that in Python ?

Up to now, I accessed all attribute as in the Python example aforementioned.

Valery
  • 25
  • 5
  • 1
    Sounds like an XY problem and like instead of a class you should use plain dict and access its keys. Unless you're not telling us what's the real issue is? Some concepts do get moved easily between languages, some don't. What is your "structure" supposed to represent? – Gameplay Mar 07 '23 at 09:02

1 Answers1

2

As I mentioned what you are trying to do is a very bad idea imho... But if you really want it you can define __getitem__ method and return value if it's in let's say set of values of your interest and raise StopIteration otherwise. This will allow to loop over those attributes using for loop and it won't return neither methods nor other attributes of that object.

class Example:
    def __init__(self):
        self.a = 0
        self.b = 1
        self.c = 2
        self.other_stuff = [1,None,False]
        self.stuff = {"a":2,"b":3,"c":87}


    def __getitem__(self, index):
        key = list(self.__dict__.keys())[index]
        if key in {"a", "b", "c"}:
            return self.__dict__[key]
        else:
            raise StopIteration()

    def foo(self):pass

e = Example()

for item in e:
    print(f"{item=}")

print(e.__dict__)

Side notes and why you should NOT do that:

  • it's unclear what's the purpose of this and it's gonna cause a headache sooner than later
  • proper data structures should be used such as dicts, dataclasses and so on
  • the dictionaries have guaranteed insert order (python 3.7+ see https://stackoverflow.com/a/60775590/15923186), so when the __dict__ is created you have to define those attributes you want to loop through in the very specific order, otherwise the StopIteration will be raised too early and you won't get all the values you want. Which is again very bad, cause it's an implementation detail and anyone using your code doesn't have to know that, neither it's obvious from looking at your code.

To see what I mean just change the order and define Example like this and again see the output of e.__dict__:

class Example:
    def __init__(self):
        self.stuff = {"a":2,"b":3,"c":87}
        self.a = 0
        self.b = 1
        self.c = 2
        self.other_stuff = [1,None,False]
...
...
print(e.__dict__)

The code won't work anymore with that change

Gameplay
  • 1,142
  • 1
  • 4
  • 16