Fonte: mail from Dan Bader PythonTricks
Hey Hans R Zimermann,
I got some questions about Python’s “lambda” expressions: What theyâre good for, when you should use them, and when itâs best to avoid them.
So I decided to write a little tutorial for you:
The lambda
 keyword in Python provides a shortcut for declaring small anonymous functions. Lambda functions behave just like regular functions declared with the def
 keyword. They can be used whenever function objects are required.
For example, this is how youâd define a simple lambda function carrying out an addition:
>>> add = lambda x, y: x + y >>> add(5, 3) 8
You could declare the same add
 function with the def
 keyword:
>>> def add(x, y): ... return x + y >>> add(5, 3) 8
Now you might be wondering: Why the big fuss about lambdas? If theyâre just a slightly more terse version of declaring functions with def
, whatâs the big deal?
Take a look at the following example and keep the words function expression in your head while you do that:
>>> (lambda x, y: x + y)(5, 3) 8
Okay, what happened here? I just used lambda
 to define an âaddâ function inline and then immediately called it with the arguments 5
 and 3
.
Conceptually the lambda expression lambda x, y: x + y
 is the same as declaring a function with def
, just written inline. The difference is I didnât bind it to a name like add
 before I used it. I simply stated the expression I wanted to compute and then immediately evaluated it by calling it like a regular function.
Before you move on, you might want to play with the previous code example a little to really let the meaning of it sink in. I still remember this took me a while to wrap my head around. So donât worry about spending a few minutes in an interpreter session.
Thereâs another syntactic difference between lambdas and regular function definitions: Lambda functions are restricted to a single expression. This means a lambda function canât use statements or annotationsânot even a returnÂ
statement.
How do you return values from lambdas then? Executing a lambda function evaluates its expression and then automatically returns its result. So thereâs always an implicit return statement. Thatâs why some people refer to lambdas as single expression functions.
Lambdas You Can Use
When should you use lambda functions in your code? Technically, any time youâre expected to supply a function object you can use a lambda expression. And because a lambda expression can be anonymous, you donât even need to assign it to a name.
This can provide a handy and âunbureaucraticâ shortcut to defining a function in Python. My most frequent use case for lambdas is writing short and concise key funcs for sorting iterables by an alternate key:
>>> sorted(range(-5, 6), key=lambda x: x ** 2) [0, -1, 1, -2, 2, -3, 3, -4, 4, -5, 5]
Like regular nested functions, lambdas also work as lexical closures.
Whatâs a lexical closure? Just a fancy name for a function that remembers the values from the enclosing lexical scope even when the program flow is no longer in that scope. Hereâs a (fairly academic) example to illustrate the idea:
>>> def make_adder(n): ... return lambda x: x + n >>> plus_3 = make_adder(3) >>> plus_5 = make_adder(5) >>> plus_3(4) 7 >>> plus_5(4) 9
In the above example the x + n
 lambda can still access the value of n
 even though it was defined in the make_adder
 function (the enclosing scope).
Sometimes, using a lambda function instead of a nested function declared with def
 can express oneâs intent more clearly. But to be honest this isnât a common occurrenceâat least in the kind of code that I like to write.
But Maybe You ShouldnâtâŠ
Now on the one hand Iâm hoping this tutorial got you interested in exploring Pythonâs lambda functions. On the other hand I feel like itâs time to put up another caveat: Lambda functions should be used sparingly and with extraordinary care.
I know I wrote my fair share of code using lambdas that looked âcoolâ but was actually a liability for me and my coworkers. If youâre tempted to use a lambda spend a few seconds (or minutes) to think if this is really the cleanest and most maintainable way to achieve the desired result.
For example, doing something like this to save two lines of code is just silly. Sure, it technically works and itâs a nice enough âtrickâ. But itâs also going to confuse the next gal or guy having to ship a bugfix under a tight deadline:
# Harmful: >>> class Car: ... rev = lambda self: print('Wroom!') ... crash = lambda self: print('Boom!') >>> my_car = Car() >>> my_car.crash() 'Boom!'
I feel similarly about complicated map()
 or filter()
 constructs using lambdas. Usually itâs much cleaner to go with a list comprehension or generator expression:
# Harmful: >>> list(filter(lambda x: x % 2 == 0, range(16))) [0, 2, 4, 6, 8, 10, 12, 14] # Better: >>> [x for x in range(16) if x % 2 == 0] [0, 2, 4, 6, 8, 10, 12, 14]
If you find yourself doing anything remotely complex with a lambda expression, consider defining a real function with a proper name instead.
Saving a few keystrokes wonât matter in the long run. Your colleagues (and your future self) will appreciate clean and readable code more than terse wizardry.
Things to Remember
- Lambda functions are single-expression functions that are not necessarily bound to a name (anonymous).
- Lambda functions canât use regular Python statements and always include an implicitÂ
return
 statement. - Always ask yourself: Would using a regular (named) function or a list/generator expression offer more clarity?
â Dan Bader