wiki:python/lambda

Python Doc

lambda functions that take a tuple instead of multiple parameters:
In Python 2, you could define anonymous lambda functions which took multiple parameters by defining the function as taking a tuple with a specific number of items. In effect, Python 2 would “unpack” the tuple into named arguments, which you could then reference (by name) within the lambda function. In Python 3, you can still pass a tuple to a lambda function, but the Python interpreter will not unpack the tuple into named arguments. Instead, you will need to reference each argument by its positional index.

Notes Python 2 Python 3
lambda (x,): x + f(x) lambda x1: x1[0] + f(x1[0])
lambda (x, y): x + f(y) lambda x_y: x_y[0] + f(x_y[1])
lambda (x, (y, z)): x + y + z lambda x_y_z: x_y_z[0] + x_y_z[1][0] + x_y_z[1][1]
lambda x, y, z: x + y + z unchanged
  1. If you had defined a lambda function that took a tuple of one item, in Python 3 that would become a lambda with references to x1[0]. The name x1 is autogenerated by the 2to3 script, based on the named arguments in the original tuple.
  2. A lambda function with a two-item tuple (x, y) gets converted to x_y with positional arguments x_y[0] and x_y[1].
  3. The 2to3 script can even handle lambda functions with nested tuples of named arguments. The resulting Python 3 code is a bit unreadable, but it works the same as the old code did in Python 2.
  4. You can define lambda functions that take multiple arguments. Without parentheses around the arguments, Python 2 just treats it as a lambda function with multiple arguments; within the lambda function, you simply reference the arguments by name, just like any other function. This syntax still works in Python 3.
Last modified 6 years ago Last modified on Feb 7, 2013, 10:12:13 AM