Python and __str__() method - bad practice or not?

Question

I've run into certain code snippet which recreates some objects every time in the __str__ method of a class.

From what I recall, the __str__ method shouldn't bother with object creation. It should just do simple formatting operations and return a string.

But I do not have any evidence for such statement. Is there any convention or perhaps a PEP explaining proper __str__ usage? I've found only PEPs regarding str vs. repr usage.

I've searched Stack Overflow, but I haven't found an answer.

Edit:

You've asked for example. Unfortunately I can't share the snippet, because it isn't mine. I'm asking for general guidelines by the community (if such thing exists).

Here's a roughly similar usage:

from collections import OrderedDict


class A:

    def __init__(self, a, b, c):

        self.a = a
        self.b = b
        self.c = c

    def __str__(self):
        # recreated every time...
        dict_sorted = OrderedDict(sorted(self.__dict__.items()))
        result = []
        for i in dict_sorted.keys():
            result.append(str((i, dict_sorted[i])))
        return ' '.join(result)

if __name__ == "__main__":
    a = A('foo', 'bar', 'baz')
    print(a)

Answer 1

It is totally okay, to create objects in the __str__ method. Python is a object oriented language, that creates objects all the time. In your example, a OrderedDict , a list, a key-iterator, some tuples and strings are created. Would you also have doubts with this __str__ -method?

 def __str__(self):
     return ' '.join(str(item) for item in sorted(self.__dict__.items()))

It does exactly the same as your __str__ -method, but creates a few methods less. There are no restrictions to a __str__ -method, other than to return a string object. But the general advice for programmers holds true: a method shouldn't do something unexpected: so __str__ should create a string representation and nothing else.