Is it possible to generate a Python function with arguments in runtime?

Question

Say,

I have a python function as following:

def ooxx(**kwargs):
 doSomething()
 for something in cool:
 yield something

I would like to provide another function with named arguments for hints as following:

def asdf(arg1, arg2, arg3=1):
 frame = inspect.currentframe()
 args, _, _, values = inspect.getargvalues(frame)
 kwargs = dict((key, values[key]) for key in args) # convert args list into dictionary form
 return list(ooxx(**kwargs))

Is it possible to have some sort of methods to generate automatically the function "asdf"? I have lots of dynamic generated ooxx functions and I would like to have corresponding asdf functions with customized named arguments. Not sure if this is the correct requirement or right way to coding :p

• jathanism: Your example code would be easier to follow if you used more meaningful variable names.

Answer 1

Your descriptions doesn't make such sense to me: You wrote a really verbose function that does this:

def asdf(arg1, arg2, arg3=1):
 return list(ooxx(**locals()))

but you want to inspect the ooxx and somehow make up appropriate names for asdfs arguments? That is impossible, there is no information about this on ooxx.

If you actually have a signature and want to create a function from it you would have to resort to eval or generate function definitions to a Python file and import it.

There is also the decorator module. You can create a function with it like this:

import decorator
asdf = decorator.FunctionMaker.create(
 'asdf(arg1, arg2, arg3)', # signature
 'return ooxx(**locals())', # function body
 {'ooxx' : ooxx}, # context for the function
 ('arg3', 1)) # default arguments

• Drake Guan: Nope. I have already prepared a function generator to generate ooxx1, ooxx2, ooxx3, ..., ooxxN. And I am looking for some sort of function generator to generate asdf1, asdf2, asdf3, ..., asdfN. For each asdfI, the number and name of arguments are different.