Is there a way to combine two decorators into one new decorator in python?
I realize I can just apply multiple decorators to a function, but I was curious as to whet
If the decorators don't take additional arguments, you could use
def compose(f, g):
return lambda x: f(g(x))
combined_decorator = compose(decorator1, decorator2)
Now
@combined_decorator
def f():
pass
will be equivalent to
@decorator1
@decorator2
def f():
pass
And to extend @Jochen's answer:
import click
def composed(*decs):
def deco(f):
for dec in reversed(decs):
f = dec(f)
return f
return deco
def click_multi(func):
return composed(
click.option('--xxx', is_flag=True, help='Some X help'),
click.option('--zzz', is_flag=True, help='Some Z help')
)(func)
@click_multi
def some_command(**args):
pass
In this example you can compose a new decorator that contains multiple decorators.
Decorators are just functions that take a function as input and return a new function. This:
@deco
def foo():
...
Is equivalent to this:
def foo():
...
foo = deco(foo)
In other words, the decorated function (foo) is passed as an argument to the decorator, and then foo is replaced with the return value of the decorator. Equipped with this knowledge, it's easy to write a decorator that combines two other decorators:
def merged_decorator(func):
return decorator2(decorator1(func))
# now both of these function definitions are equivalent:
@decorator2
@decorator1
def foo():
...
@merged_decorator
def foo():
...
It gets a little trickier if the decorators accept arguments, like these two:
@deco_with_args2(bar='bar')
@deco_with_args1('baz')
def foo():
...
You might wonder how these decorators are even implemented. It's actually pretty simple: deco_with_args1 and deco_with_args2 are functions that return another function decorator. Decorators with arguments are essentially decorator factories. The equivalent of this:
@deco_with_args('baz')
def foo():
...
Is this:
def foo():
...
real_decorator = deco_with_args('baz')
foo = real_decorator(foo)
In order to make a decorator that accepts arguments and then applies two other decorators, we have to implement our own decorator factory:
def merged_decorator_with_args(bar, baz):
# pass the arguments to the decorator factories and
# obtain the actual decorators
deco2 = deco_with_args2(bar=bar)
deco1 = deco_with_args1(baz)
# create a function decorator that applies the two
# decorators we just created
def real_decorator(func):
return deco2(deco1(func))
return real_decorator
This decorator can then be used like this:
@merged_decorator_with_args('bar', 'baz')
def foo():
...
If you don't want to repeat yourself too much in a test suite, you could do like this::
def apply_patches(func):
@functools.wraps(func)
@mock.patch('foo.settings.USE_FAKE_CONNECTION', False)
@mock.patch('foo.settings.DATABASE_URI', 'li://foo')
@mock.patch('foo.connection.api.Session.post', autospec=True)
def _(*args, **kwargs):
return func(*args, **kwargs)
return _
now you can use that in your test suite instead of a crazy amount of decorators above each function::
def ChuckNorrisCase(unittest.TestCase):
@apply_patches
def test_chuck_pwns_none(self):
self.assertTrue(None)
A bit more general:
def composed(*decs):
def deco(f):
for dec in reversed(decs):
f = dec(f)
return f
return deco
Then
@composed(dec1, dec2)
def some(f):
pass
is equivalent to
@dec1
@dec2
def some(f):
pass
Yes. See the definition of a decorator, here.
Something like this should work:
def multiple_decorators(func):
return decorator1(decorator2(func))
@multiple_decorators
def foo(): pass