从同一个类中调用类成员变量会在Python中产生NameError

Question

I'm trying to call a class member variable from within its class, but I get a NameError: name '...' is not defined .

A similar situation is created by means of the following minimum working example:

from pprint import pprint


class MyClass:

    _my_class_variable = {'key_0': 0,
                          'key_1': 1}
    _my_keys = _my_class_variable.keys()

    pprint(_my_class_variable)  # WORKS!
    pprint([value for value in _my_class_variable.values()])  # WORKS!
    pprint([_my_class_variable[key] for key in _my_keys])  # DOES NOT WORK!
    pprint([_my_class_variable[key] for key in _my_class_variable.keys()])  # DOES NOT WORK!

which returns NameError: name '_my_class_variable' is not defined .

How is it possible that the first two pprint command work but not the last two pprint commands?

Answer 1

Everything in the list comprehension is run in a separate scope (as a function, basically), except for the iterable used for the loop. So, on the lines that don't work, _my_class_variable is not defined.

One way to solve this would be to make sure that you pass _my_class_variable to an immediately executing lambda function so that it's available in the list comprehension's scope :

pprint((lambda _my_class_variable : [_my_class_variable[key] for key in _my_class_variable.keys()])(_my_class_variable))  # DOES WORK!
# or
pprint((lambda _my_class_variable=_my_class_variable : [_my_class_variable[key] for key in _my_class_variable.keys()])())  # DOES WORK!

Why does this work? List comprehensions get their own separate scope in Python 3 and are wrapped in a temporary function object and called immediately. If you were to create an explicit scope for the _my_class_variable , like in a function, its scope will be considered when resolving _my_class_variable .