1109
votes

I want to check if a variable exists. Now I'm doing something like this:

try:
   myVar
except NameError:
   # Do something.

Are there other ways without exceptions?

14
What's wrong with the exception?S.Lott
@S.Lott: if myVar is something really complicated, that takes a long time to produce/evaluate, wouldn't the try slow things down?dbliss
@dbliss: It's a variable. Aside from some really weird cases if you're doing something crazy with exec or metaclasses, it's not going to be expensive.user2357112 supports Monica
A more complete answer: stackoverflow.com/a/1592578/1661797nickboldt

14 Answers

1883
votes

To check the existence of a local variable:

if 'myVar' in locals():
  # myVar exists.

To check the existence of a global variable:

if 'myVar' in globals():
  # myVar exists.

To check if an object has an attribute:

if hasattr(obj, 'attr_name'):
  # obj.attr_name exists.
138
votes

The use of variables that have yet to been defined or set (implicitly or explicitly) is often a bad thing in any language, since it tends to indicate that the logic of the program hasn't been thought through properly, and is likely to result in unpredictable behaviour.

If you need to do it in Python, the following trick, which is similar to yours, will ensure that a variable has some value before use:

try:
    myVar
except NameError:
    myVar = None      # or some other default value.

# Now you're free to use myVar without Python complaining.

However, I'm still not convinced that's a good idea - in my opinion, you should try to refactor your code so that this situation does not occur.

By way of an example, the following code was given below in a comment, to allow line drawing from a previous point to the current point:

if last:
    draw(last, current);
last = current

In the case where last has not been bound to a value, that won't help in Python at all since even the checking of last will raise an exception. A better idea would be to ensure last does have a value, one that can be used to decide whether or not it is valid. That would be something like:

last = None

# some time passes ...

if last is not None:
    draw(last, current);
last = current

That ensures the variable exists and that you only use it if it's valid for what you need it for. This is what I assume the if last was meant to do in the comment code (but didn't), and you can still add the code to force this if you have no control over the initial setting of the variable, using the exception method above:

# Variable 'last' may or may not be bound to a value at this point.

try:
    last
except NameError:
    last = None

# It will always now be bound to a value at this point.

if last is not None:
    draw(last, current);
last = current
74
votes

A simple way is to initialize it at first saying myVar = None

Then later on:

if myVar is not None:
    # Do something
26
votes

Using try/except is the best way to test for a variable's existence. But there's almost certainly a better way of doing whatever it is you're doing than setting/testing global variables.

For example, if you want to initialize a module-level variable the first time you call some function, you're better off with code something like this:

my_variable = None

def InitMyVariable():
  global my_variable
  if my_variable is None:
    my_variable = ...
20
votes

for objects/modules, you can also

'var' in dir(obj)

For example,

>>> class Something(object):
...     pass
...
>>> c = Something()
>>> c.a = 1
>>> 'a' in dir(c)
True
>>> 'b' in dir(c)
False
11
votes

I will assume that the test is going to be used in a function, similar to user97370's answer. I don't like that answer because it pollutes the global namespace. One way to fix it is to use a class instead:

class InitMyVariable(object):
  my_variable = None

def __call__(self):
  if self.my_variable is None:
   self.my_variable = ...

I don't like this, because it complicates the code and opens up questions such as, should this confirm to the Singleton programming pattern? Fortunately, Python has allowed functions to have attributes for a while, which gives us this simple solution:

def InitMyVariable():
  if InitMyVariable.my_variable is None:
    InitMyVariable.my_variable = ...
InitMyVariable.my_variable = None
9
votes

catch is called except in Python. other than that it's fine for such simple cases. There's the AttributeError that can be used to check if an object has an attribute.

6
votes

A way that often works well for handling this kind of situation is to not explicitly check if the variable exists but just go ahead and wrap the first usage of the possibly non-existing variable in a try/except NameError:

# Search for entry.
for x in y:
  if x == 3:
    found = x

# Work with found entry.
try:
  print('Found: {0}'.format(found))
except NameError:
  print('Not found')
else:
  # Handle rest of Found case here
  ...
5
votes

I created a custom function.

def exists(var):
     return var in globals()

Then the call the function like follows replacing variable_name with the variable you want to check:

exists("variable_name")

Will return True or False

0
votes

Short variant:

my_var = some_value if 'my_var' not in globals() else my_var:
0
votes

This was my scenario:

for i in generate_numbers():
    do_something(i)
# Use the last i.

I can’t easily determine the length of the iterable, and that means that i may or may not exist depending on whether the iterable produces an empty sequence.

If I want to use the last i of the iterable (an i that doesn’t exist for an empty sequence) I can do one of two things:

i = None  # Declare the variable.
for i in generate_numbers():
    do_something(i)
use_last(i)

or

for i in generate_numbers():
    do_something(i)
try:
    use_last(i)
except UnboundLocalError:
    pass  # i didn’t exist because sequence was empty.

The first solution may be problematic because I can’t tell (depending on the sequence values) whether i was the last element. The second solution is more accurate in that respect.

0
votes

Like so:

def no(var):
    "give var as a string (quote it like 'var')"
    assert(var not in vars())
    assert(var not in globals())
    assert(var not in vars(__builtins__))
    import keyword
    assert(var not in keyword.kwlist)

Then later:

no('foo')
foo = ....

If your new variable foo is not safe to use, you'll get an AssertionError exception which will point to the line that failed, and then you will know better. Here is the obvious contrived self-reference:

no('no')

---------------------------------------------------------------------------
AssertionError                            Traceback (most recent call last)
<ipython-input-88-d14ecc6b025a> in <module>
----> 1 no('no')

<ipython-input-86-888a9df72be0> in no(var)
      2     "give var as a string (quote it)"
      3     assert( var not in vars())
----> 4     assert( var not in globals())
      5     assert( var not in vars(__builtins__))
      6     import keyword

AssertionError: 
0
votes

Also a possibility for objects, use __dict__.

class A(object):
    def __init__(self):
        self.m = 1

a = A()
assert "m" in a.__dict__
assert "k" not in a.__dict__
0
votes

It may not be performant, but you generalise the solution to a function that checks both local variables and global variables.

import inspect
def exists_var(var_name):
    frame = inspect.currentframe()
    try:
        return var_name in frame.f_back.f_locals or var_name in globals()
    finally:
        del frame

Then you can use it like this:

exists_var('myVar')