Access a function variable outside the function without using "global"
Question:
I am trying to access a local function variable outside the function in Python.
I can make code like this work with global variables:
bye = ''
def hi():
global bye
bye = 5
sigh = 10
hi()
print(bye)
Next, I tried this code, hoping to access bye outside hi() without using global bye:
def hi():
bye = 5
sigh = 10
return
hi()
x = hi()
print(x.bye)
This gives AttributeError: 'NoneType' object has no attribute 'bye'.
Next, I tried:
def hi():
bye = 5
sigh = 10
return bye
hi()
x = hi()
print(x.bye)
This didn’t improve matters; I get AttributeError: 'int' object has no attribute 'bye'.
Is there a way to access a local function variable (bye) outside its function (hi()) without using globals and also without printing out the sigh variable? How can I do it?
Answers:
The problem is you were calling print(x.bye) after you set x as a string. When you run x = hi() it runs hi() and sets the value of x to 5 (the value of bye; it does NOT set the value of x as a reference to the bye variable itself). EX: bye = 5; x = bye; bye = 4; print(x) prints 5, not 4.
Also, you don’t have to run hi() twice, just run x = hi(), not hi(); x=hi() (the way you had it it was running hi(), not doing anything with the resulting value of 5, and then rerunning the same hi() and saving the value of 5 to the x variable.
So full code should be
def hi():
bye = 5
sigh = 10
return bye
x = hi()
print(x)
If you wanted to return multiple variables, one option would be to use a list, or dictionary, depending on what you need. For example:
def hi():
return { 'bye': 5, 'sigh': 10 }
x = hi()
print x['bye']
You could do something along these lines (which worked in both Python v2.7.17 and v3.8.1 when I tested it/them):
def hi():
# other code...
hi.bye = 42 # Create function attribute.
sigh = 10
hi()
print(hi.bye) # -> 42
Functions are objects in Python and can have arbitrary attributes assigned to them.
If you’re going to be doing this kind of thing often, you could implement something more generic by creating a function decorator that adds a this argument to each call to the decorated function.
This additional argument will give functions a way to reference themselves without needing to explicitly embed (hardcode) their name into the rest of the definition and is similar to the instance argument that class methods automatically receive as their first argument which is usually named self — I picked something different to avoid confusion, but like the self argument, it can be named whatever you wish.
Here’s an example of that approach:
def add_this_arg(func):
def wrapped(*args, **kwargs):
return func(wrapped, *args, **kwargs)
return wrapped
@add_this_arg
def hi(this, that):
# other code...
this.bye = 2 * that # Create function attribute.
sigh = 10
hi(21)
print(hi.bye) # -> 42
Note
This doesn’t work for class methods. Just use the instance argument, named self by convention, that’s already passed to methods instead of the method’s name. You can reference class-level attributes through type(self). See Function’s attributes when in a class.
You could do something along this lines:
def static_example():
if not hasattr(static_example, "static_var"):
static_example.static_var = 0
static_example.static_var += 1
return static_example.static_var
print static_example()
print static_example()
print static_example()
def hi():
bye = 5
return bye
print hi()
To be able to access a local function’s variable, one might add the name of the function and a dot before the name of the local variable (and then, of course, use this construction for calling the variable both in the function’s body and outside of it). This solution works in Python 3.7.4.
For example:
def func(): # define a function
# here y is a local variable, which I want to access; func.y
# defines a method for my example function which will allow me to
# access function's local variable y
func.y = 4
x = func.y + 8
return x
func() # now I'm calling the function
a = func.y # I put its local variable into my new variable
print(a) # and print my new variable
If you want to avoid global, one possible approach is to define a class. Each class instance has its own attributes; there is also a class attribute space where instances can share an attribute between them.
Object-oriented programming can be challenging to get into if you are new to Python, but this might actually be a good time to start playing with it.
class Thing:
shared = "foo"
def __init__(self):
"""
This gets called when you create a new Thing()
"""
self.bar = "baz" # default value for new instances
def get_bar(self):
return self.bar
def set_bar(self, value):
self.bar = value
Now, let’s create two instances.
first = Thing()
second = Thing()
The get_bar and set_bar methods are not strictly necessary in simple examples like this one. You can also do
second.bar = "ick"
print(second.bar)
# "ick"
print(first.bar)
# "baz"
(though for more complex scenarios, you probably want to require users to call the setter and getter methods; there are ways to force this – see e.g. What's the pythonic way to use getters and setters?)
If you change a class attribute via one instance, it will not be changed in the other instances, either.
second.shared = "poo"
print(first.shared)
# "foo"
But if you change it in the class itself, it will be changed in all the instances which have not separately overridden the shared value.
Thing.shared = "zoom"
print(first.shared)
# "zoom"
print(second.shared)
# "poo", still
To recap, you create a new Thing instance by calling Thing(); this will run the __init__ method before returning the new instance. Inside the class, the instance is the first argument to every (non-static, non-class) method, and conventionally called self (though you could get away with calling it shirley if you wanted to, as far as the Python interpreter is concerned).
There’s a lot more to classes; the main selling point is probably that you can create subclasses which inherit from their parent class but can override some behaviors (common examples often involve real-world concepts like animals or vehicles, but a class can just be anything where you want to create a type and encapsulate its behavior, and perhaps override some methods in derived types).
I am trying to access a local function variable outside the function in Python.
I can make code like this work with global variables:
bye = ''
def hi():
global bye
bye = 5
sigh = 10
hi()
print(bye)
Next, I tried this code, hoping to access bye outside hi() without using global bye:
def hi():
bye = 5
sigh = 10
return
hi()
x = hi()
print(x.bye)
This gives AttributeError: 'NoneType' object has no attribute 'bye'.
Next, I tried:
def hi():
bye = 5
sigh = 10
return bye
hi()
x = hi()
print(x.bye)
This didn’t improve matters; I get AttributeError: 'int' object has no attribute 'bye'.
Is there a way to access a local function variable (bye) outside its function (hi()) without using globals and also without printing out the sigh variable? How can I do it?
The problem is you were calling print(x.bye) after you set x as a string. When you run x = hi() it runs hi() and sets the value of x to 5 (the value of bye; it does NOT set the value of x as a reference to the bye variable itself). EX: bye = 5; x = bye; bye = 4; print(x) prints 5, not 4.
Also, you don’t have to run hi() twice, just run x = hi(), not hi(); x=hi() (the way you had it it was running hi(), not doing anything with the resulting value of 5, and then rerunning the same hi() and saving the value of 5 to the x variable.
So full code should be
def hi():
bye = 5
sigh = 10
return bye
x = hi()
print(x)
If you wanted to return multiple variables, one option would be to use a list, or dictionary, depending on what you need. For example:
def hi():
return { 'bye': 5, 'sigh': 10 }
x = hi()
print x['bye']
You could do something along these lines (which worked in both Python v2.7.17 and v3.8.1 when I tested it/them):
def hi():
# other code...
hi.bye = 42 # Create function attribute.
sigh = 10
hi()
print(hi.bye) # -> 42
Functions are objects in Python and can have arbitrary attributes assigned to them.
If you’re going to be doing this kind of thing often, you could implement something more generic by creating a function decorator that adds a this argument to each call to the decorated function.
This additional argument will give functions a way to reference themselves without needing to explicitly embed (hardcode) their name into the rest of the definition and is similar to the instance argument that class methods automatically receive as their first argument which is usually named self — I picked something different to avoid confusion, but like the self argument, it can be named whatever you wish.
Here’s an example of that approach:
def add_this_arg(func):
def wrapped(*args, **kwargs):
return func(wrapped, *args, **kwargs)
return wrapped
@add_this_arg
def hi(this, that):
# other code...
this.bye = 2 * that # Create function attribute.
sigh = 10
hi(21)
print(hi.bye) # -> 42
Note
This doesn’t work for class methods. Just use the instance argument, named self by convention, that’s already passed to methods instead of the method’s name. You can reference class-level attributes through type(self). See Function’s attributes when in a class.
You could do something along this lines:
def static_example():
if not hasattr(static_example, "static_var"):
static_example.static_var = 0
static_example.static_var += 1
return static_example.static_var
print static_example()
print static_example()
print static_example()
def hi():
bye = 5
return bye
print hi()
To be able to access a local function’s variable, one might add the name of the function and a dot before the name of the local variable (and then, of course, use this construction for calling the variable both in the function’s body and outside of it). This solution works in Python 3.7.4.
For example:
def func(): # define a function
# here y is a local variable, which I want to access; func.y
# defines a method for my example function which will allow me to
# access function's local variable y
func.y = 4
x = func.y + 8
return x
func() # now I'm calling the function
a = func.y # I put its local variable into my new variable
print(a) # and print my new variable
If you want to avoid global, one possible approach is to define a class. Each class instance has its own attributes; there is also a class attribute space where instances can share an attribute between them.
Object-oriented programming can be challenging to get into if you are new to Python, but this might actually be a good time to start playing with it.
class Thing:
shared = "foo"
def __init__(self):
"""
This gets called when you create a new Thing()
"""
self.bar = "baz" # default value for new instances
def get_bar(self):
return self.bar
def set_bar(self, value):
self.bar = value
Now, let’s create two instances.
first = Thing()
second = Thing()
The get_bar and set_bar methods are not strictly necessary in simple examples like this one. You can also do
second.bar = "ick"
print(second.bar)
# "ick"
print(first.bar)
# "baz"
(though for more complex scenarios, you probably want to require users to call the setter and getter methods; there are ways to force this – see e.g. What's the pythonic way to use getters and setters?)
If you change a class attribute via one instance, it will not be changed in the other instances, either.
second.shared = "poo"
print(first.shared)
# "foo"
But if you change it in the class itself, it will be changed in all the instances which have not separately overridden the shared value.
Thing.shared = "zoom"
print(first.shared)
# "zoom"
print(second.shared)
# "poo", still
To recap, you create a new Thing instance by calling Thing(); this will run the __init__ method before returning the new instance. Inside the class, the instance is the first argument to every (non-static, non-class) method, and conventionally called self (though you could get away with calling it shirley if you wanted to, as far as the Python interpreter is concerned).
There’s a lot more to classes; the main selling point is probably that you can create subclasses which inherit from their parent class but can override some behaviors (common examples often involve real-world concepts like animals or vehicles, but a class can just be anything where you want to create a type and encapsulate its behavior, and perhaps override some methods in derived types).